Collectible Card Game Headquarters

Wow, ty for this!

Observations and questions, if I may

1) Why so many attempts at the black castle?

2) It's not that I don't like mind control as such, it's that I can't bloody see if the actual deck is working or not XD Four out of 5 matches vs. the priestess were mind control in my last run.

3) Warlock stats are indicative, but it's to be expected - it was build with a completely different philosophy than most others on purpose. Need to change it though.

4) On the same note, all tier 3 enemies (except current Enchantress) are a too steep step above (most) tier 2 enemies right now. Heck, some are a steep step above tier 4 enemies.

5) 9 - 1 vs. Crusader?

6) 2 - 0 vs. Sorcerer?

7) How come you have so many matches vs Mind Stealer, and how many of those were against his actual deck?

I like the score vs. Sedge Beast.

9) 19 - 3 vs. Tusk Guardian - I thought he'd be tougher.

10) Did you get enough of a chance to play vs. the Vampire Lord and get properly locked out by his combos? I know what he's doing wrong, but when he draws right, currently, he doesn't look like the essence of it needs fixing.

11) 14-2 vs. Centaur Shaman, how come so many?

12) How was Lord of Fate? I'm worried about that one.

13) How was Mandurang?

14) 4 - 0 Shapeshifter - that guy needs attention.

15) Arzakon needs more work, deffinitely, I just stuck some domain stuff in there at random times.

16) What do you think of the scores and the whole package overall currently? What numbers do you think I should be shooting for?

17) Are most of these very high number of duels vs. particular opponents dungeon scores? Those can seriously skew things in the players favor.

lujo wrote:Wow, ty for this!

Observations and questions, if I may

1) Why so many attempts at the black castle?

2) It's not that I don't like mind control as such, it's that I can't bloody see if the actual deck is working or not XD Four out of 5 matches vs. the priestess were mind control in my last run.

3) Warlock stats are indicative, but it's to be expected - it was build with a completely different philosophy than most others on purpose. Need to change it though.

4) On the same note, all tier 3 enemies (except current Enchantress) are a too steep step above (most) tier 2 enemies right now. Heck, some are a steep step above tier 4 enemies.

5) 9 - 1 vs. Crusader?

6) 2 - 0 vs. Sorcerer?

7) How come you have so many matches vs Mind Stealer, and how many of those were against his actual deck?

I like the score vs. Sedge Beast.

9) 19 - 3 vs. Tusk Guardian - I thought he'd be tougher.

10) Did you get enough of a chance to play vs. the Vampire Lord and get properly locked out by his combos? I know what he's doing wrong, but when he draws right, currently, he doesn't look like the essence of it needs fixing.

11) 14-2 vs. Centaur Shaman, how come so many?

12) How was Lord of Fate? I'm worried about that one.

13) How was Mandurang?

14) 4 - 0 Shapeshifter - that guy needs attention.

15) Arzakon needs more work, deffinitely, I just stuck some domain stuff in there at random times.

16) What do you think of the scores and the whole package overall currently? What numbers do you think I should be shooting for?

17) Are most of these very high number of duels vs. particular opponents dungeon scores? Those can seriously skew things in the players favor.

Oh boy, that's a load of questions!

1) They kicked my butt! So many times. Hypnotic Specter from the start is nasty! I didn't have much luck with the dungeon layouts either - many enemies to fight, not a huge help from dice, starting life going down as the black wizard took over more cities (he had 3 when I finally won, and there were a few more that the other wizards had held that I never reclaimed. I had 28 starting life in the end). Also, there are so many cards in the game I do not know. This means I know the answer to maybe 20% of the riddles, so those aren't easy shortcuts like they were for me in the original game. Plus most of the black decks are quite strong. I scored losses to all of the different ones here.

2) Ah, for playtesting resons. Good point.

5) Well, most of my critters are white (Crusade and Jihad help them too), and I have board wipe and good fliers. And a fast creature deck when it works. First turn Birds of Paradise, second turn Qasali Pridemage, third turn Armadillo Cloak, attack for 5 and steal as much life. Open defense so AI attacks, which means I can keep on attacking. Even if the AI has blockers, I have a bigger critter, with Trample.

6) Don't remember how those duels played out, really. But he only has 12 life, so a good start from me and he's in trouble.

7) I tried to get duplicate cards (he's one of only two creatures with this ability that you do not need to be on the narrow strips of coastline to meet) and wasn't succesful for a long time. Mind Control less than half the time, IIRC.

8. Me too.

9) Once I got a defense up, he was pretty hamless. I had evasion or board wipe eventually, and lots of life gain when I could attack.

10) Yes, he dominated those three duels he won.

11) If those numbers count the duels in the castle, then it's because he was the easiest enemy for me in there, so I picked him when I had otherwise equal choices when trying to clear a way to the black wizard.

12) Pretty nasty with his Pestilence a few times. I remember it as a solid and interesting deck to play against.

13) He's the one with the Hondens, right? It was...different. Only two matches, though. On the one I won, he had two Hondens out for a long time, the red one and the discard one, but that mattered little because my hand was already empty and I had a large creature out that stole more than enough life.

14) Probably.

15) There are two interesting alternatives in the overhaul thread. One specializes in strong multicoloured creatures and one is a sliver deck (but lacking the sliver cards from Nomad's Bazaar and maybe even from Thieves' Hideout).

16) No idea what to aim for. It was a fun experience, at least this first time. Would be a lot harder to play less obviously strong decks in this environment.

17) I didn't visit that many dungeons, really, and in most of them, didn't need many duels to get what I wanted.

Thanks a ton for this, v. important info, all of it! Some comments:

1) Yes, I've noticed that the black castle is much tougher than other castles, too - but before I added any decks! Hypnotic Specter on the board for free is probably the best example of something that's completely messed up and should never be done. Folks might say that they want the game tough, but noone really wants it that tough or that kind of tough.

5) vs. Crusader. Ah, well, that explains it. If you get the advantages from his crusades and can deal with longbow archers (and have strong lifegain), he's probably not that tough, if you don't (and most decks don't), he's way too murderous IMO.

12) I'm pleasantly surprised about Lord of Fate, occasionally I think I messed that deck up, and occasionally he kicks my ass mercilessly. I think he has a bit of the same problem as Vampire Lord - too uneven depending on the draw, need to look into that.

13) vs. Mandurang - hmmm, I seem to have undertested vs. lifelink, because I just didn't give too much lifelink to enemies. He was impossibly tough in the first version, so I nerfed him (you played vs. a nerfed version). I think a few other decks you found you can handle have the same... well, it's not a problem really, but lets say feature. They don't handle lifelink the same way they handle other stuff. I have just one Rhox War Monk in my deck currently, it's dominating everything and winning me matches on his own.

14) I think I could make him work fine enough with domain, it's just that I just stuck a few cards in here and there and not actually reworked him yet. But now that you mention it - one of the things that gives the player an edge in general is the fact that I didn't go crazy with multicolored creatures and a lot of those tend to be significantly above the curve compared to monocolored creatures. A lot of matches I've been winning currently are obviously due to 2 and 3 colored bombs from multicolored blocks which naturally outclass the opposition.

16) Heh, I hear you - you've been using lifelink, I've been using some other gimmicks, but the thing is that most of the enemy decks aren't actually strong decks. Or rather, when they're strong decks they're not strong because I thought it was a good idea, but because the AI couldn't handle more balanced cards right.

And when they just seem to be (significantly) tougher than the OG decks that's because the OG decks were mostly quite disfunctional. A deck with a proper curve and able to play all it's cards somewhat right (still constrained by the AI) is going to look like someone upped the power level and difficulty significantly simply because all of a sudden things work with no gimmicks.

That's why I'm going on about Shandalar design decisions being on the whole quite unwise. The game got away with it to a degree because the the decks and the AI weren't effectively there, most of it was, so to speak, "painted on cardboard", not functional at all. So what was going on is that the player would be playing a sealed deck against decks which didn't really work (for a variety of reasons) and that whole aspect didn't look like it could be the basis of a game. But when enemy decks just work, when they can beat you on a regular basis, when they're a significant enough factor, just the proposition of having to build up a sealed deck into something that can take on constructed decks is visibly enough of a challenge (as any experienced mtg player could tell you). So it's easy to notice how much you have to lean on exploiting AI blind spots to actually win, and what a difference certain mechanics make, for example. Before, the AI was at a disadvantage because it couldn't play MtG, now the player is at a disadvantage because the core proposition, one of starting with a sealed deck vs even decentish constructed decks, is in-and-off-itself a rather brutal one. Even if those decks were made by a very experienced deckbuilder trying to gimp them enough to lessen that.

In other words, OG Shandalar worked because it's core mechanic, dueling, didn't actually work. The "strongest" decks in Shandalar seemed strong simply because those were the few decks the AI could play anywhere close to right, or they had cards in them which are obscenely powerful (or brutal in the format) - the actual deck never worked, but the Strip Mines , Mishra's Factories , Sol Ring s , Mana Vault s, Serra Angel s and whathaveyous would win matches for it creating an illusion that the decks work when they didn't.

The current Crusader actually works - it has no evasion, no direct damage, hardly any rares, cards a mtg veteran would scoff at, but try beating it at 10 life with a non-white deck. Heck, the Troll Shaman missplays most of it's cards , in human hands the deck is significantly more deadly and yet this deck works X00% better than most OG Shandalar decks. And it's not more powerful than many of those decks on paper, it's just that it actually manages to work while they never did. The current Tusk Guardian is a much more powerful deck - but the AI misplays it so throughly that it looks balanced or weak - it's not, it's just non-functional There's no landwalk in the deck pool now, very little protection from color, no moxes or sol rings or things made when folks making MtG didn't have a good grasp on what breaks their game, and yet the game is, overall, quite a bit more difficult and takes hand-picking obviously strong stuff or stuff the AI can't react to at all to be properly playable.

So if the inherent challenges of Shandalar are (get a pile of cards into something that can take on constructed decks) + (do this while losing cards every time you lose a match) + (do this vs. an AI that can only really use powerful cards properly) + (do this starting at half as much life as the game was balanced for) + (more) , that's just too much. (Start with a sealed deck) is the main thing, can't get rid of that, (AI) is also a tall order so you can't get rid of that, (Ante, in just the card gain and loss on wins and losses) also has to be there...

But asking for more than that is just asking to juggle more balls than can realistically be juggled. I suppose it just takes some playing with a pack of functional decks to be able to see what I mean more fully. Adding more cards and being able to make decks the AI can play at all (4realz) really pushes out many things Shandalar developers needed/got away with because they couldn't make their core mechanic (MtG) work right.

Here's my current progress and impressions:

Enemies | Open

Witch 2/1
Undead Knight 5/0 (Really ought to push him back up a bit)
Warlock 4/4 (Yep, this guy is too low for Corrupt & Pillar Tombs spam)
Vampire Lord 3/1 (I'll touch him up after the next update)
Nether Fiend 1/1 (If I move Pillar Tombs of Aku over and chase Master of the Feast out, and get Abhorrent Overlord to play the Role of Corrupt...)
Necromancer - 0 (Dunno why I'm avoiding this guy...)

Cleric - 1/1
Priestess - 2/0
Crusader - 2/4
Paladin - 3/1
Arch Angel - 1/3 (See? The deck has AI issues, though, it's very prone to attacking with it's protection guys. They'd mess me up even worse because all my bombs are multicolor, though. It also spams mass removal madly, but otherwise it's flying vigilant beatdown... When Arch Angels besiege something I stopped using the sword - getting there later and fighting something else is a much better deal)
High Priest - 4/1 (Gramps somehow has issues. Not entirely sure what the deal is yet)

Seer - 1/2
Merfolk Shaman - 5/1 (I'm not even unhappy with him, could use just a bit more bite)
Conjurer - 0/1
Sea Dragon - 1/0 (There's only one real problem with this guy, it's that he tends to throw down his big splashing bounce first and then put serpents down, and that he's misusing Phantasmal Terrain, otherwise he'd be really strong)
Shapeshifter - 1/4 (This guy has a bug, and it's Body Double destroying all permanents. Otherwise - don't even thing playing against him with a 40 card deck. That happens. In that sense, mission accomplished, now just to refine him a bit. Made me switch over to a 60 card deck.)
Thought Invoker - 1/0 (The misplays are strong with this one. Also it had the "AI skips playing stuff" bug going on, so findings are inconclusive, sample size too small)

Druid - 6/1 (Dungeon!) (Just one Rhox War Monk in the deck very early on ran a dungeon over and got me a ton of wins vs. green guys because I had a trillion life)
Elvish Magi - 4/1 (Dungeon!)
Enchantress - 6/2 (Dungeon!) (Funnily enough she can still kick your ass if she draws cards I made too difficult to draw because it used to be able to kick my ass every time with every hand it drew.)
Forest Dragon - 4/2 (Beat me at 100 life in dungeon. This one was a really lousy deck originally, cool idea terrible execution, and currently it suffers a lot from it's regenerating snake being bugged to not be able to regenerate, the reach snake not being in the pool, it playing lures on wrong things and several other things. Needs work, but it would be wise to do this work in the next patch. Very disfunctional.)
Beastmaster - 3/2 (Has .csv issues with some things, has issues with
Summoner - 0/1

Sorceress - 3/0 (rework already under way)
Sorcerer - 1/2 (you think you have him, give him one wee little error, and you don't. He actually has .csv issues with some of his spells, wouldn't you know it)
Troll Shaman - 2/0 (I have a lot of bounce. Well, a lot of AEther Bursts. Almost had me that second time with a 10/10 hellion)
Goblin Lord - 0/1 (This guy is getting a deck with no lords in it.)
Hydra - 1/0 (Needs a bit of rethinking)
War Mage - 1/0

Tusk Guardian - 1/0 (He's awfully misplayed in multiple ways. It's a bit strange because he's 4 pumps, 4 equipment and a bunch of dudes, but there you have it)
Sedge Beast - 1/0
Ape Lord - 2/0 (His good hand is really strong, but I made it happen too irregularly)
Centaur Shaman - 1/0
Winged Stallion - 2/1
Fungus Master - 1/2 (I'm avoiding this guy, he's too damned explosive)
Centaur Warchief - 2/0
Mind Stealer - 3/0
Lord of Fate - 4/0 (Got a better idea about what he needs done now, the deck is pulling in too many directions currently).

Aga Galneer - 0/1 (Oh god, brutally ran me over)
Saltem Tor - 2/0

Kiska-Ra - 3/2 (Was interesting, but Pandora's Box wasn't there too often or working out for him when it was. Madly cheap weenies, madly good pump and most of all Control Magic did the work when he won. I could actually make him quite a bit stronger I think).

What with all the bugs and everything this may sound strange, but while looking for a workable land base formula for multicolored opponents I got around to putting together a new Ape Lord. It's being misplayed and Kird Chieftain is waiting for the next update to be fixed, and some .csv and targeting tweaks would probably help it (although you never know), but I'm going to be rolling with it to see how it goes:

New Ape Lord | Open

.1710 4 Stomping Ground
.10460 4 Rootbound Crag
.91 8 Forest
.164 8 Mountain

I'll put in a few words on this mana base in it's own paragraph below as well as write about land bases a bit.

Let's get vanilla stuff out of the way first

.8752 2 Brute Force - This was various things at various points.
.12962 4 Ground Assault - Spitting Earth Mk. II .
.9548 3 Gloomwidow - I kinda felt the deck needed reach at one point.

.437 3 Kird Ape - It's just silly ole early beats
.14125 4 Kird Chieftain - He's bugged atm but fixed in the next update.

Now:

.11297 4 Liquimetal Coating - THIS is what makes it special. Also looks like an ape is on the art.

It combines with:

.2493 3 Goblin Tinkerer - In theory could kill all enemy lands.
.1698 3 Uktabi Orangutan - In theory it killes stuff. In practice it gets played without a target with 2 Liquimetal Coating in hand.
.7914 2 Nullmage Shepherd - A "paladin" variant, for resolving stalls.
.1610 4 Indrik Stomphowler - In theory a beefy Uktabi Oranguntan . In practice 3 get played in a row with no targets with any number of Liquimetal Coating sitting in hand. Yes, really.

.4935 2 Artifact Mutation - This is, in theory, nuts. Also in practice, and combos with Nullmage Shepard, but in practice it can sit uselessly in hand as things get played over Liquimetal Coating.
.14612 2 Return to the Earth - This is, in practice, nuts.*

This isn't the only way to build it, there are several problems with it, but considering all the possible problems you could have while trying to build a RG deck in Shandalar it seemed to play somewhat well and was at least interesting in testing. Missplayed, but interesting.

The general idea is that it can perform regular beatdown. It's a bit slow, and sometimes mishandles it's lands but it can work. Once Kird Chieftain becomes able to pump stuff, if that doesn't misfire spectacularly it'll be much better at it.

Brute Force was Temur Battle Rage for a longish while, and that certainly made the beatdown plan quite explosive (and very play-card-kill-opponent). That seemed to cause the deck to skip playing Liquimetal Coating, play all the cip-kill-an-artifact things with no targets, and then end up with just Liquimetal Coating in hand and play it last for massive booing. Brute Force, I think, leaves more chances for it to stick that massively important thing down at the right time.

Another rather good thing that I think is messing up the Liquimetal Coating plan is the Ground Assault. You can't replace it, it's madly good, it's just that the AI tends to play it and then just skip the coating and do a cascade of everything that combos with the coating without the coating for no reason. It's possible that something about that is going to have to be done eventually.

The choice of Spider is also problematic - that is, it's not, it's perfect, it's just that it can ALSO cause the coating to not get played when it needs to be played. There's a combination where I ditch the Nullmage Shepherd and Gloomwidow for Towering Indrik, clear some other problematic choices for Darksteel Axe and that might end up working, except it's a significantly less optimal plan with it's own issues.

Otherwise - the short descriptions tell the story. Liquimetal Coating + artifact hate is the name of the game, and it can mean a ton of Vindicate all over the place whenever the AI pulls it's s**t togather. Several hours went into just getting it to here, no matter how simple and striaghtforward the concept might look at first glance.

I hope it works in-Shandalar better than it worked in infinite-duels mode. There's a good chance it works better than the old ape lord simply because I added the basic "boost any deck" things - reach, Return to the Earth , cost-efficient targeted removal, theoretically better mana base and all that. I think I'll be trying to develop this one as the default no-csv-editing-needed one as it looks like it might actually work and have a theme and all that. Or not, but the biggest issues might be that the AI is just terminally incapable of playing any mana base other than alpha duals.

* Return to the Earth - this card is seriously starting to bug me. You make a deck which has a way to use any artifact removal to kill anything, but the deck only really takes off when you add a card which can pretty much just kill anything on it's own. Easy way to make this deck work better? Take out everything that combos with Liquimetal Coating, and just replace with these. Then do the same in every green deck. You get a Vindicate, and you get a Vindicate, and you get a Vindicate... -.- Sadly, I think I'll end up having to do something like that in the end, at least with this one. Problem is - if there's a Liquimetal Coating out, this gets tossed at a land. And the Goblin Tinkerer doesn't seem to be able to target lands

I think the whole Distorting Lens - Northern Paladin - Southern Paladin - Liquimetal Coating and all that jazz needs to be looked into in regards to what can and what can't target lands atm , do we want it to be able to target lands , and what can be activated when. It's a big ole mess.

Feedback much appreciated.

About the potential mana bases:

Land Bases | Open

So, what I tried with the Ape Lord, and what I'm going to try with all 2 colored opponents (with a possible exception here and there), will be the routine Shockland (like Stomping Ground up there) + Checkland (like Rootbound Crag up there).

Lands that come into play tapped generally tend to mess the AI up. What's more, lands that come into play tapped under certain conditions make the AI wildly unpredictable in practice. This means that predicting how a deck is going to work out with those lands in it is quite a bit more difficult than if it had other kinds of lands. This is why I've generally gone with the Painland (like Karplusan Forest) + alpha dual (like Taiga ), as neither comes into play tapped.

That plan has to go one way or another because alpha duals are too damned good, and the vast majority of MtG hasn't been designed/developed/tested/played with them in mind at all. In fact, when shocklands (which also have basic land types but cip tapped unless you pay 2 life) came around about 10 years into the lifespan of mtg they were an enormous power / flexibility boost compared to the lands mtg was built around before them. So you can imagine how good alpa duals really are.

Shockland + Checkland is available in all colors, and it allows the AI a high chance of having all it's lands come into play untapped for a low-to-moderate life cost. They also obviously play well with each other. There are other options for other purposes, but this one is so much superior to all the other ones that I really, really, really hope it works. If the AI can only really work with alpha duals we're up a creek.

Another thing is that hardly anything compares to alpha duals when it comes to playing with fetchlands, and fetchlands are key in many strategies. Shocklands aren't bad with them but the Shandalar life shennanigans make them significantly worse than they are IRL. The life bussiness also compounds problems with finding satisfying mana bases for the AI. Often, the best lands cost life, and if you want a reliable mana base that the AI can be trusted with this life loss tends to add up.

The unwanted byproduct of saddling the AI with shock + check lands is that you have to boost it up to compensate for land-mishandling screwups. And the problem with that is that people pay hundreds of dollars IRL to be "saddled" with those lands, so what happens is that I might have to give the AI the best
cards to go with the best-short-of-plainly-busted lands.

What to do with the 3-color opponents I'm not too sure. Checklands get worse if there's not enough basics in a deck, and the tri-colored guys don't have their problems solved by dual lands, so they have to run enormous ammounts of non-basices. Tri-colored lands are good for them, but they can only allow themselves to run those because they run 12 alpha duals and fetches on the side. You can't give them Shocklands because the ammount of Shocklands they would need would either kill them or just make everything they have come into play tapped additionally slowing them down. And even if you gave the AI all City of Brass with no damage it would still screw up non-land related things, making it's life any more difficult is pushing your luck hard.

But let's see if this works for the 2 color guys for now and leave that for later. From what I've seen from 2 color guys, it's going to take a huge time investment to get 3 colored guys right or even reach a solid conslusion that they just can't work without alpha duals. I've pretty much completely neglected fine-tuning mana bases so far as it's a pain to do that irl, and here you've got the AI and the "random-card-missing-entirely-lol" ante thing screwing with you on top of that. To make matters worse, nothing is missing in infinite-duel, but then your fetch targets or whathave you go awol in regular shandalar.

I get the feeling people think working decks just pop into existance and that deckbuilders have Time Walk recursion going on IRL. Or that you can just peel a decklist of the web, eyeball aproximate substitutions for missing cards based on "meh, it's from the same ccg", and call it a day. In reality you can't even be sure a full-vanilla deck will be played in a sane manner even after you have pages of notes of what insane behavior you can expect and cards and templates to stay away from flagged in a separate deckbuilding program. And then the otherwise right choice of removal causes the AI to loop into a spiral which makes it not play the combo-cornerstone before it's played all the cards which play off it. But if you give it a different piece of removal something else causes a cascade of insanity, or if it doesn't the deck suddenly has glaring mana issues, or after many iterations you find out it's impossible for the AI to pull off something simple because of tangential reasons and you have to scrape the whole thing, repeat ad nauseum...

AI has Mountain, Kird Ape, Rootbound Crag, Liquimetal Coating and Uktabi Orangutan in hand. AI plays Rootbound Crag turn one, Mountain turn 2, plays Kird Ape, turn 3 it plays Uktabi Orangutan with no target. It does that. No sane player would. The difference in effect between that and what any player would do? No potentially 2/3 creature on the board turn 1, Uktabi Oranguntan is a 2/2 instead of a Vindicate, and Liquimetal Coating gets played well after it's already played all it's Indrik Stomphowler s, it has a dead card in hand, it has 1-several dead cards on the board. AI dies to fliers. Which would all be dead otherwise as it would be spamming Vindicates every turn. What happened? It played its lands in banally stupid fashion. Imagine trying to deal with this as just a deckbuilder, then with this when it takes juggling 3 colors. That's mana bases for you. Plays wrong land just once - deck fails horribly, looks like a toothless mess, anyone using it figures a random pile cards put together by a 10 year old is better. Plays the right land in a way any 10 year old would - deck murders you with an endless stream of "destroy target permanent", folks either cheer because the deck is awesome or rightfully worry that it's too damned brutal. In an alarming number of situations it does the wrong thing or happens to be set to always do the wrong thing (probably for unrelated reasons).

And also that's deckbuilding, deckbuilding is mostly hours and hours and hours of that. And what I hate the most is knowing I would have to report this if I wanted it potentially fixed, but wth do I report, exactly? How to I explain it to someone busy with something else, probably equally or even more complicated? Just picking Rootbound Crag over a Mountain on turn 1 can and will cause a clusterf**k of insanity, and if there isn't particular AI for Rootbound Crag then making the AI behave sanely with it could cause a million other lands to malfunction, and if it were to be done right someone would have to go rewrite checklands specifically, and... Eh. It's not that it's a thankless job (or even a job at all XD), but it's ugly more often than not.

---

Did some more testing instead of uploading - that deck doesn't work and I've reported 3 things:

1) AI is unable to play checklands. It prioritizes them over basic lands and timewalks the oponnent, then spends the next turn playing it's 1st turn play, then just keeps trainwrecking the decks. This is really awful.

2) Goblin Tinkerer never seems to target lands. That's all he's in the deck for, and this never happens, but what does happen is:

3) You can't make a Liquimetal Coating deck simply because the AI can't get it into play. It can use it fine enough, but if there is anything at all it could play instead, no matter how insane, it will do so. Second Goblin Tinkerer with no targets on turn 3? It will do it. Uktabi Orangutan with no targets, with 2 Goblin Tinkerers on board who could blast something if it played the coating? It will play the Orangutan as a vanilla 2/2 for 3. 3rd Indrik Stomphowler with no targets in a row while being pounded by fliers? It will do it. Liquimetal Coating in hand. When it works it only works by accident because it happened to not have anything to play but the coating on turn 2 or 3.

Only way it would work is if I simply took out all (and I mean all) the cards in the deck which cost 1-4 mana out of the deck, made everything cost 5 mana except the coating and then pray it draws the coating before it gets to 5 mana. This is so obviously insane, but that's what you'd have to do currently to get the AI to just put a 2 mana artifact into play, even though more than half of it's deck is only in the deck because of that artifact and it can even use it right once it's in play.

Back to the drawing board, but if this happens to get addressed there's a really cool Ape Lord deck ready to happen.

Hey Lujo,
Do you have a link to the latest set of decks you have?
Also, are you using Nomad's Bazaar 1 or Thieves Hideout 3?

How do you test your decks? Is there an easy way to playtest your decks vs. Vanilla or some of the decks Rodil made here


Thanks,

Well, I can easily upload the folder right now, but since both me and jiansonz just went through a campaign (he did a full campaign I did 2/3 of one) I need to apply quite a bunch of tweaks first. Feedback is an awesome thing ^^

They're Nomads Bazaar 1, and as soon as the next update hits they'll be that because Korath has fixed quite a number of bugs in the meantime (some of which caused me to not use some stuff so far). Also because a only some percentage of the cards are actually usable by the AI, so more cards = more cards to work with (also helps bugtest, ofc).

I generally first test them in infinite duels mode which is what Korath made possible particularly for deck testing and bug catching. What you do is you go start->run and in the bar there you put this:

C:\Magic\Manalink3\Program\Shandalar -p decks -e decks/0437.dck

For you it would be whatever your main shandalar folder is. "- p decks" means that the player plays a random deck from the "decks" folder which is where the shandalar decks are. "-e decks/0437.dck" means the enemy always plays the deck called 0437.dck (which is Ape Lord, which I've been testing recently).

You can make a folder called Vanilla, put vanilla decks in them and write "-p vanilla" and then the player would get a random vanilla deck. Same for enemies or whatever. If it's unclear, ask away.

I mainly test them like that just to see if the deck is doing what it's supposed to be doing at all, while at the same time getting the feel of how the other (random) decks from the pool handle in player hands to check for screwups and possibly bugs.

But doing that isn't necessarily what would produce any particular results - the targeted strength of a deck is a very relative thing. If my deck beats a vanilla deck or gets owned by a rodil deck doesn't say a lot about which of those is more appropriate for the actual game.

What you can do, and this was always helpful, is give me a day or two or several to catch up with the data so far to tweak the decks that we nailed as in need of tweaks, then just start a campaign. I didn't want to contaminate the forums with another build, but I can bundle my .exe and a few other files with the spirit tweaks which make black ones play right in my current witch deck, and which is also hacked to make conjurer and troll shaman visible (I'll also take off Mind Control so that enemies that have it will be using their actual decks).*

One more thing is is that I can also bundle my ini with it with infect critters taken out of the pool, adjusted starting cards in dungeons and debug mode enabled. That one is easy to do on your own, it allows you to make the enemy hand visible so that you can see what it's doing, that's very handy in infinite duel mode. It's impossible to see just how much the AI is screwing something up without that, and it can also be used to adjust lives on the fly (and do other things).

EDIT: This is unnecessary, but helpful. Just playing a campaign with my ini and the deck pack and making notes leisurely along the way whenever something catches your eye, displeases you or whatever is great. Also win/loss totals vs. enemies, those I love, they can tell a lot.

Good news everyone!

No, really! Plumbing the depths of error cascades wasn't the only thing I was doing, and all the pool mapping and Forge editing I've done seems to have payed off. There's three for a start with notes, and more are coming

CAVEAT: These were "only" tested in infinite duels mode for "does the AI do what it's supposed to be doing" and "do the errors look tolerable or potentially beneficial". Power level fine-tuning and scaling will take someone actually plugging them into a campaign (but I do have an unfinished one going, I just thought it'd be pointless to keep going despite conclusive evidence these need work).

Ape Lord! Yes, really! | Open

So instead of diverging from the old tweaked one in an interesting direction which needs card AI tweaks applied, I went in the "boring-but-more-solid" direction. Dunno about power level, but it does have it's charms, I hope.

.1710 4 Stomping Ground - Shockland
.1994 4 Karplusan Forest - Painland
.91 9 Forest
.164 9 Mountain

There really wasn't anything else that the AI could be handed. This is a strictly worse option than the currently non-working one with checklands. That's because checklands are something you can feed the player, and which the player can appreciate, while painlands don't really mix that well with the Shandalar life rules. At the same time neither lets player easily break multicolored bombs and run 5 color whatever. So this is the choice you'll be seeing a bunch when the deck in question cares about it's curve and there isn't a very specific reason to add a third set of duals.

It's 26 lands which might be a bit much and cause mana flood here and there, but the deck is somewhat top heavy and was having issues with curving out properly during testing before I bumped it.

.12962 3 Ground Assault - Removal that doesn't get wasted on the player.

.8752 3 Brute Force - Well, it's still Shandalar and the AI still won't budge if you don't give it a Giant Growth. I'm not giving it Colossal Might that easily, though.

.437 4 Kird Ape - Stooge no1
.14125 4 Kird Chieftain - Stooge no2
.4379 4 Silverback Ape - Stooge no3

.12359 2 Eaten by Spiders - Meh, I'm not giving him Return to the Earth just like that. He can have this and like it.
.9548 4 Gloomwidow - I did give him this. I took reach out of the pool too hard on the initial pass, was sick of spiders. Gloomwiddow is at least interesting. Let's see how it goes, too much quality reach tends to make beatdown too well-rounded.

.14246 2 Crater's Claws - To be honest, this thing is probably both unnecessary and not all that useful in this deck from what testing has shown.
.14281 4 Heir of the Wilds - But still, I went with the boring option, so some flavor isn't bad.

.13495 2 Nylea, God of the Hunt - HAH! Well, Kird Chieftain is still missflagged and his ability doesn't work, the 1G Enchantment from Exodus (name escapes me) isn't in the pool, so this seems to be the only actual way to give stuff mass trample. I don't expect it to be a dude all that often, and there's potential for it to mess up with the pump, that needs more testing.

.10209 2 Soul's Majesty - Well, I can't give everyone Harmonize and I specifically can't give this guy Shamanic Revelation. If this gets played on Nylea it's pretty lolsy, there's other fine targets, too.

There we go, that's not so much a deck as much as putting functional red and green cards around some very no-nonsense apes, with one big risk in the god. Lots of way to get very annoyed at it, but it's really mostly just blunt trauma. Which is a shame, RG is surprisingly deep with many memorable builds including honest-to-god control*, it's just that most of it doesn't work or is even stronger than this and there's Djinns and Dragons to save functional stuff for too. *The thing about RG control is that noone realizes that what they're playing against is actually a control deck, which tends to trip people up all the time. This isn't one, this is bog standard (even sub-par) beatdown.

Now then, this guy had a proper rework coming for a while:

Druid! Still shennanigany and hip! | Open

.91 20 Forest
.8962 4 Llanowar Reborn

All the hours of testing multicolored land bases thaught me that if you want to make the AI life difficult you just hand it lands which come into play tapped. So I decided to try for this with the decks which are too functional if they're just functional - the initial opponent decks!

.5756 4 Woodland Druid - Looks measly, but the deck supports him.
.5961 4 Ironshell Beetle - Druid turn 1, this turn 2, Druid swings.

.13165 4 Kraul Warrior - This is both an early bear and the end-game mana sink.
.5631 4 Nantuko Elder - And this is where the mana comes from, and helps with the cip-tapped lands (the AI tends to play those after basics here). Also a Druid!

.14119 4 Invasive Species - I took a gamble here. It can slow the deck down even further and is vulnerable to bounce and stuff, but on the other hand it can bounce Llanowar Reborn and Ironshell Beetle (this happens). This also plays minorly well with the lucky charm. Keep an eye out for this thing, I hope it's not actually unworkable, didn't seem to be during testing.

.1094 3 Nantuko Disciple - And another risk. Well, not really a risk, this guy belongs in this deck mechanically. The problem is that the AI has been all over the place with similar effects so far and this guy is actually first-pick material in odissey draft, so I hope he does work properly. If the AI isn't making your life miserable with these, it's doing something wrong.

.1481 2 Kaysa - She's a leftover from the previous deck. I'm not sure this one actually needs her, it might be strong enough as it is, but then again, she's legendary and pumps the little guys nicely. Lords and Shandalar don't really mix, but she's kinda ok. Also, I need a green legend in any green deck I can squeeze one into (Ape has the god) because anyone running green has no reason not to run Iwamori of the Open Fist who's a huge pain in the a**.

.1737 2 Wurm's Tooth - There it is. 2 Might be too few, but let's see.
.12431 2 Natural End - Some equipment proofing and lifegain in a nicely druidic package. Not giving this guy Return to the Earth either.
.1333 3 Spidersilk Armor - I did give him this. This is a silly card i keep throwing around too much.
.2863 4 Bee Sting - And ofc, Bee Sting! Because Hornet Sting just gets tossed at the enemy head.

Right, so it's an insect/druid deck with several things going on. Several risky cards too, both in terms of ways to screw up and to be too strong. Those can be addressed I think and I'd be delighted if it turned out to work. It's got character and a good flavor/identity.

There, that should be at least more functional that the old rework and much more friendly to green players trying to play against it. Feedback practically necessary. It kicked Sedge Beast's backside and also did in several other decks I didn't expect it to and came back from very dire situations, so my hopes are kinda up.

And then there's... this girl!

Sorceress, now with more of a point! | Open

.164 16 Mountain
.10835 4 Smoldering Spires - The obligatory roadblock in case it needs it. Also ties in with other things mechanically.
.1218 4 Great Furnace - There's only one reason for these in this deck (I'll give a speech at some point on metalcraft and battle cry and how the whole "mtg is about turning dudes sideways now" controversy actually came about, just not now).

.1736 2 Dragon's Claw - Lucky Charm, I'm experimenting with 2 per deck.

This is what it uses to build a bunker:

.11535 2 Strandwalker - Reach? Reach!
.10673 3 Spidersilk Net - More reach?! More reach! And equaly as important - toughness. Caveat: I've seen the AI arse up with this, once, by spaming it's equip between creatures pointlessly. I hope Korath manages to nail what's causing that behavior at some point with equipment in general. Keep an eye out if you're testing this.

.11395 3 Turn to Slag - What with all the reach, the deck can bunker down and live to see it. Doubles as equipment hate, solid removal and doesn't get thrown at the face. It was necessary not to mess up the curve in the previous deck, this one is more laid back, so it might be able to play a less expensive removal thingy. Thank god for these horrible-looking but great-at-what-they're-for cards that did somehow make it into Manalink and Shandalar!

.993 4 Cinder Wall - The roadblock. With Spidersilk net it's quite a roadblock, but thankfully not a durable one.

.10549 4 Goblin Shortcutter - This guy can, combined with the tapland, open up space for the deck to poke you instead of just bunkering down.
.10938 4 Goblin Tunneler - This guy is the main win condition, in the same vein.
.4720 4 Moggcatcher - And these guys sit behind and churn out the tuneling gobbos. That's what I messed up last time, I gave it more than "exactly the gobbos that it wants to have", so this thing went haywire.

.360 3 Sisters of the Flame - These are on flavor and can use the tunels, too. Also, provide mana for the more expensive stuff to alleviate the tapland effects, as by the time one of these is operational the deck has run out of basic lands and is playing taplands (from what I've seen)
.2881 4 Fire Imp - These would be over the top in a deck with actual burn in it, but since it's just them and Turn to Slag, they're probably fine.

So all those guys are non-evasive, non combat-worthy weenies which need the equipment strapped onto them to become combat worthy, and need Moggcatcher s to bring in the tunnelers.

.14167 3 Scrapyard Mongrel - This guy, on the other hand, doesn't. All the other artifact friendly stuff, like much of what red has in the pool, makes terrible early opponent tools. It's all just so lethal. This guy is hardly harmless (especially when equipped), but there needs to be something scary to pull aggro off the Moggcatcher and Goblin Tunneler so that they can do their thing.

It looks scary, and some of it IS scary, but it manages to be a red deck which is theoretically quite vulnerable to red decks (as hating out it's artifacts means you can pound the weenies into oblivion). It might need tweaking, it needs Shandalar testing, but it's certainly a step in the right direction from the old Sorceress.

All three of these have a bit of the same vibe as the current Sorcerer, they probably work as they're supposed to work (or close to it, or will after easy tweaks), but then they might turn out to be huge annoyances to play against.

Anyway, I've got a bunch more stuff done in Forge and waiting for infinite-duel testing. Getting to it tomorrow. I'll upload these without the whole folder, you can stick them right in if you're playing my stuff no need to restart anything. Feedback very wellcome.

Lujo,

Thanks for the info. I'll wait for you to make your tweaks and the next update.
Hope it comes out soon

At that point I'll start a new campaign and give you some feedback on what I find.

AI is the Honden of Infinite Rage | Open

Well, day (&night) completely wasted trying to make a Centaur Warchief. It's uncanny, whatever I put together in RW just falls apart. Even if I just stick cards that worked in other decks in it ends up sucking, being criminally misplayed or, most often, both. At this point I'd settle for "player has exactly 0 chance against this deck, it will always beat him but it at least it does what it was meant to do".

I've just had the AI hold Heliod, God of the Sun in it's hand and play lands for 4 turns going from 3 to 7 mana. I'm mana screwed, I got nothing, it plays Heliod it's GG. It waits forever, does nothing, is mana flooded, that Heliod would've totally rocked, its the only thing in it's hand, no reason whatsoever not to play him... I figure, must be that "do nothing all game" bug or something. No, at seven mana, for no apparent reason, it plays Heliod.

I'm left thinking - is it capable of this sort of insanity with other gods? Or just Heliod? Does this mean I have to go take other gods out of decks? Wth do I think?

Game before that it has 4 mana on board and a Sacred Foundry in hand and 20 life. It has Serra Angel in hand. It puts Sacred Foundry down, tapped. Why? So it can cast Brute Force on something which wasn't blocked for insignificant damage instead of playing the Serra Angel.

How can people go "AI generally plays things fine"? How? The Brute Force is in the deck for no other reason but because without it it will never attack with anything that can be blocked. That alone is enough cause for anyone who'd say "AI generally plays things fine" to drop dead of shame on the spot - mtg players don't need Brute Force in every deck to attack. But then, because it's in the deck, that can cause the AI to not play a Serra Angel just so that it can throw the Brute Force at the opponents head.

And this isn't a random play, this will always happen. You can substitute Brute Force for hundreds of other cards, and you can substitute Serra Angel for hundreds of other cards and the AI will always, like clockwork, screw it up.

Folks, does anyone have a RW decklist that they believe works in Shandalar? It doesn't matter if it turns out it only looks like it works if you squint and turn your head the other way, I'd simply try anything at all at this point.

---

The situation with Beastmaster is even more mindblowing. I made a deck with Garruk's Companion, Kalonian Tusker and Leatherback Baloth. And gave it pack hunt so that it can always draw 4. Just to see if it can screw it up, those are 3 vanilla creatures with insane stats, in mono green. It can't play it. If you put up up enough ground defense to be able to kill any one of them in combat, it will just sit there stacking them.

This means you can have 5 power total on the board and hold off an army of those. I want to make a video of it, put it in my sig, and every time someone says something good about the AI post a link to that. Read those three cards again, carefully. It can't play them properly. Those cards. The AI needs decks to be specifically built so that it can use those cards. You need to devote deck space to increase it's chance of being able to use THOSE right.

Just think about that.

I can imagine that it would take inhumane effort to fix that, though. That level of hopeless means something rather deep is rather deeply wrong.

Same question - does anyone have a really tough mono green beatdown decklist which they think works but doesn't have Overrun or lords in it? I won't critique it or anything I just want to give something else a whirl and see what happens. It's either that or physically banging my head against the wall until I cause enough brain damage to really start thinking like the AI properly.

I actually sat thinking about the last sentence up there. I've been up all night and day, I was really light-headed and too tired to actually pay attention to anything I was doing anymore and felt legitimately stupid. So I gave the RW thing one more shot... and it worked! For a given value of worked. If this turns out well, I suppose I'll just get drunk next time or something...

New Centaur Warchief, I suppose:

Centaur Warchief, the 6AM edition | Open

.1708 4 Sacred Foundry
.5334 4 Battlefield Forge
.188 8 Plains
.164 8 Mountain

The mana base. It will probably cause cascades of bad decisions, but if you play without seeing the AI's hand you probably won't notice the extent of it.

.229 3 Smoke - Err, the last build had something going with this, left it in.
.12895 3 Blind Obedience - Also this. I think the extort part is causing a lot of stupidity to happen.
.13454 2 Heliod, God of the Sun - Because mass vigilance. The painful stupidity is strong with this one, though. (EDIT: did some more testing, it can get very cringeworthy, I'm worried about it being displeasing)

.10928 4 Flame Slash - Because I'd have to really lose my mind before I give it Swords to Plowshares.
.1160 2 Star Compass - Because... Because I had to give it stuff that costs a bunch to play and this might randomly get it to play it faster. Maybe. Also it might help to pay for stuff. (EDIT: Lol, I meant activation costs)

Right, so, the plan:

.6627 4 Bonesplitter - The painfully plain-looking but way-too-good equipment from the first ever set with equipment in it. You see me giving this to the AI you know I'm really desperate. (EDIT: Here's a pro tip, if you want to push a deck over the top in an inconspicious way, stick these in.)
.7557 3 Shuko - This was Bone Saw until five minutes ago when I decided to see if not having to pay for equip would make it play a bit more logically. Jury's still out on that. It might decide to just stay out of it.

Anyway, it's supposed to take those things and stick them onto these things:

.6698 3 Leonin Den-Guard
.10789 4 Kitesail Apprentice
.1235 3 Skyknight Legionnaire

In the manner of the few decks it could get anywhere with in OG Shandalar, which is to say sticking Unholy Strenghts onto way-too-good weenies. These aren't exactly way too good (they're no slouches, mind you), but the buffs are soooo much better than the old ones.

.9087 2 Brion Stoutarm - Then if it has a stroke of genious it can actually chuck them at you to gain life. That's supposed to be the plan to resolve inevitable air stalls. Not that it can't just beat you up with this guy. (EDIT: The combination of this guy, Heliod and Bonesplitters can lead to so many moments which will have you staring at the screen in disbelief. Not in a good way.)
.9734 3 Archon of Justice - Oh, and this was Serra Angel. In the real world this deck neither wants or needs a 5 drop, but in the bizzaro world, it does. Serra is more fool-proof because it's pretty much the only card the AI can use as intended in the whole game, and also naturally plays well with Smoke. This is so much more annoying, and is a Serra Angel if Heliod is on the board. (EDIT: After some testing, this actually plays kinda fine in and off itself. I don't really expect the AI to be able to capitalize on Brion and this, I just wanted something that punished enemy removal in some way.)

Right, so the deck has 3 glaring flaws with no AI madness a) it sucks vs removal as there's no card draw and too much non creature-stuff and b) it's draws will be wildly uneven because there's too much non creature stuff and recovering from a bad hand is harder because there is too few creatures and c) too much stuff is too expensive and it has no cheap card draw to really help. You can also add d) the AI is a f**tard know to misplay just about every card in the deck, often in a variety of ways.

Not too much I can do about any of that. Maybe. I'm open for suggestions, if I already tried them I can at least share what what went wrong or rethink it in a new light.

What it's got going for it is that the distastefully strong equipment and the not-half-bad weenies can mess anyone up, Smoke + Blind Obedience can slow you down to a crawl, and the bombs are pretty bomby.

No real need to test it for errors, unless you want to punish me for being a nuisance and call me an incompetent hack for putting so many faulty cards into an uneven pile of a deck. Which is fair game, IMO, I got no honest defense against that, fire away.

But what I would appreciate feedback on is whether this pile is too overwhelming. Stupid as that sounds it can be really deadly, and I wouldn't even think of putting some of this stuff in a deck if I felt like I had a choice.

Well, it's probably better than the last one, it really needed a rework. Got a bunch more to go through, Thought Invoker just flat out needs .csv work ( Mana Drain , of all things, can't work unless I adjust everything in the deck I give it to to be playable before combat ) or it'll be as much of a pain, and Beastmaster might even be doable but I might really need to get drunk first.

Enjoy!

Oh, and here's the new Fungus Master. The problem with the current one was that it was left half-unfinished. It was somewhat left in an early-alpha state where I just pile up a bunch of stuff to spot what the AI is likely to spontaneously use, so I cleaned it up. The other thing was that, well, that it was too explosive. It still is, actually I haven't really fixed that for several reasons, but this version is much more wortwhile to develop (I think).

No major groaning, either, apart from one minor thing I reported way back which has been reviewed, if you pick the right cards for this sort of thing (trial and error) the mechanics happen to (mostly) perfectly coincide with what the AI tries to do with anything. It also looks like someone took care with some aspects of it, too.

Fungus Master | Open

.126 8 Island
.91 8 Forest
.5457 4 Yavimaya Coast
.1715 4 Breeding Pool

The new standard land package.

And here's the creature curve:

.12909 4 Cloudfin Raptor - Grows large and flies.
.12970 4 Gyre Sage - Grows large, helps play stuff, is very kick-ass.

.8316 4 Vigean Hydropon - Main man of the deck. He's pretty much an enchantment (Afiya Grove), but he evolves the little guys, grafts the big guys bigger, and does various other things. Ideally he'd also enable fight cards, but those aren't in yet. What he doesn't do is attack or block.

.1251 4 Carven Caryatid - A late addition, because I kicked Fungusaur and Rootwater Hunter out, and many other things weren't working out. The deck needed card draw, and this evolves the little guys while also being rock-solid defense. Possibly too solid. Embodiment of Spring was a bit too passive (and misused - the deck would play it before the other little guys, and did several other things wrong). It also made the deck have too many unimpactful draws. Wall of Blossoms, though, is a legit alternative and I'll try it in the next round of tweaks.

.1365 4 Assault Zeppelid - Whoo, boy. If these guys follow a turn 3 Hydropon they're not only huge cheap filers but they also evolve the little guys. Too good, the alternatives were unsatisfying. If the deck spends the first three turns playing inoffensive looking stuff (including a Hydropon which doesn't actually do anything), it really need something to stabilize. It might need axing, though.
.14312 2 Longshot Squad - These guys were a late addition, but I'm actually happy with them. Spike Soldier uses it's self-pump in a way which instantly kills it, but gets played as if it won't (AI not very good or sensible with spikes in general, except...)

.3429 3 Spike Colony - this one. This one it can play, it just keeps the train going, and mostly ends up serving as a walking pump spell, but that's what I wanted it there for anyway.

.13031 2 Prime Speaker Zegana - And she tops the curve off and refills the hand. Crazy strong card.

What does it all mean, there's a lot of text on those cards and it all looks unfamiliar, and how come the AI can navigate this and it can't handle Leatherback Baloth. Simple, Evolve and Graft take a lot of words to explain, but are in fact really simple. If the AI just naively plays it's cards as the lands come along, the mechanic plays itself. It's been sorta well coded to avoid obvious pitfalls (I did find some exceptions to this, but meh, good job, more than fine), and all the mechanics involved had enough cards ported that I could find the ones among them that don't cause the AI to screw it up and at the same time comprise a deck which can win.

Little guys grow, big guys also grow, there's 8 X fliers that are/can get big and a lot of pump which ensures AI attacks. It's so effortlessly explosive most of the time that I tried to keep it down by making it unable to deal with permanents, and gave it:

.11394 3 Turn Aside - Which is supposed to protect it's creatures while they grow or once they've grown.
.14387 3 Stubborn Denial - Which does mostly the same thing (also potentially counters other stuff) which can easily get it's ferocious off the hydropon and general evolve everywhere.

The AI does screw them up, though - it's too eager to play it's stuff early on which ties up it's mana so it often can't cast these. But the deck is also all lords anyway and just the chance to screw you up as you're scrambling to keep it from overwhelming you is worth it. I can't really give it anything else, either, it would have to be either Frantic Search (which it doesn't really want to play) or Snap (which it tends to blow on whatever, but might be handy, except I don't want to make it any more overwhelming).

.10993 3 Might of the Masses - This it plays fine. Meaning it does what it always does with pump, it is often braindead when you consider it but happens to work right in this deck. Also, maybe I'm hallucinating but I think I've seen it sometimes not throw it away for some extra damage on an unblocked dude, but rather keep the mana and play something after combat instead. Which is what it's supposed to be doing with every Giant Growth variant ever, except it never does that. 95% of the time you see the AI play pump on an unblocked creature a baby seal dies, and god knows what doesn't get played because of that terrible move. Whatever's causing that to be the default AI behavior, if changed to not happen, would instantly make the game several orders of magnitude better. As things are you have to give the AI cards which it is set to misplay, which cause deck-wrecking misplay cascades quite often, because if you don't give it to the AI it won't attack with non-evasive dudes.

I think... Hmmm, I think I'll check on Might of the Masses just to see if it happens to actually play it right. If anyone knows of any pump spell which the AI won't throw out all the time regardless of whether the creature is blocked or not please let me know. If you know of a decklist with it also that as it might be due to context, but that would still be a big deal. I'm also sure it doesn't take being a mtg pro to work out why this is a big deal, but if someone is for some reason curious I can try to explain. It's one of those things that make "AI plays stuff fine overall" unspeakably false, but if this problem didn't exist neither would many others and that sentence would be much truer.

Enjoy!

Tonight's progress, I'll update the post as I go along. I'll also try my best to keep it short. This sort of thing helps me think and ease frustration, and today my outlook is quite mellow: a) a lot of stuff doesn't work because things are by convention flagged to be played post combat. Knowing this lessens the sense of powerlessness, I can do something about that, and b) I've a better awareness of just how deep the whole mad pump spam can go and can more easily identify what's going wrong when something is. It might even fix itself to a degree if more stuff can be played pre-combat, who knows?

Anway, I opened it up with a light tweak session and feel a lot of gratitude for Korath for porting one particular card:

Cleric Tweaks | Open

.188 15 Plains
.8545 4 Flagstones of Trokair
.10829 4 Sejiri Steppe

The lands now resemble what I'm going for with other low-level opponents. Flagstones randomly thin the deck and help threshold and the Steppe is... well, it can't hurt. It's a tossup between it and Kabira Crossroads , I like the potential for an occasional offensive play this has.

.5533 4 Dedicated Martyr - Threshold Helper
.5628 3 Mystic Penitent - Thershold dude

.5843 4 Mystic Familiar - Threshold dude
.5629 4 Mystic Visionary - Threshold dude

.4454 3 Devout Witness - Threshold helper and equipment proofing

.5864 4 Possessed Nomad - Threshold dude. No splash, it's a legit creature at this level and it's also has a cool exorcist/cleric flavor thing going, as well as it nicely serves to show white's also a villain here.

.2799 3 Tithe - Threshold helper and grabs lands. As soon as there's another functional cantrip it'll replace this, as the deck has a rather low curve. I think I'll stick Sacred Wine back instead of this.

.11966 3 Ghoulcaller's Bell - Main threshold helper and can also randomly "ante" enemy lands or key cards as collateral.

.3246 3 Repentance - I did give him some removal. It can misplay it and it's not the most reliable removal ever, so, again, if the cantrips ever work out this should be a cantrip that pumps.

.14603 4 Pressure Point - This is what I'm grateful for currently. White threshold lives off cantrips and land sacrifice, and this is the closes to functional as I can currently find. Well, actually, that's not true, there's another thing I'm pondering, so I might revisit the deck later tonight and apply changes. For now this seemed great.

.1733 2 Angel's Feather - I added this, even though it's not exactly what the deck wants. It wants things that go to the grave and draw it cards, not things which stay on the board and don't. I'll know more after the Priestess rework later tonight.

This is just to help me see it all written down and explained and if I do end up changing it later tonight I'll note it down below, and then upload all the decks I do.

The deck is simple, put weenies on board, get threshold, everything flies, swing. I think I could make more of the non-creatures cantrips (and think that I should, except I'm not the happiest with the available cantrips), I think it's possible one of the non-cantrips could be a piece of equipment, I think sejiri steppe could be Kabira Crossroads and that Angel Feather could go to the priestess. So some of it is up in the air.

Next up - Priestess!

Priestess - the wall free edition | Open

The wall deck was and is a problematic concept. More on that in the section called "Mind Control" below. This deck is a sprit-aura based one and I've tested the concept in a lot of variants before. There are systematic misplays which are easy to get rid off by applying general spirit tweaks, which I'll do at some point. They're somewhat cloaked here by the fact that this deck is actually strong and seems quite reliable.

.188 20 Plains
.10576 4 Kabira Crossroads - This goes into this deck, Cleric gets the other one. It's easy to figure out why

.14111 3 Heliod's Pilgrim - This girl tutors for auras.
.7572 3 Tallowisp - This thing can tutor for a LOT of auras.

.7294 4 Lantern Kami - Flying Tallowisp trigger and pest.
.7264 3 Kami of Ancient Law - Handy Tallowisp trigger and pest.
.7675 2 Kataki, War's Wage - Also Handy Tallowisp trigger and pest.

.8845 4 Serra's Boon - Removal it can fetch with the tutors.
.10619 4 Nimbus Wings - Flying it can fetch with the tutors.
.12957 2 Gift of Orzhova - Flying and lifelink it can fetch with tutors. Also follows up on that unnerving presence of black in white with these churchy types for flavor and atmosphere building reasons. Can go out, though.

.13994 3 Tethmos High Priest - What it can put auras on for crazy good effect.
.11962 3 Geist-Honored Monk - Likewise and also can be crazy strong and is another priestess type to go along with Heliod's pilgrim

I'm positive those two fit into the deck both thematically and mechanically, the question is - does this make the deck too damned good? Because this sort of deck only really fits the "Priestess" out of all enemies, but when you put it together right it might be a bit on the overkill side for a tier 2 opponent.

Also:

.4180 2 Replenish - To bring auras back from the dead and alleviate their inherent risks. This is a lolworthy use of such a distinguished card, but you don't want to see a grim and serious one, believe you me. Still, some of those auras are removal.

.8831 3 Rebuff the Wicked - And this is there to try to make it difficult to remove it's stuff. I'm seeing strange plays happen with these in hand, or rather not happen. Might not be due to it though, and 3 might be too many.

Anywho - it's a compact, memorable deck with a unique identity. The question is is whether it's too strong, but that can likely be alleviated with nerfs and limiting what it has access to. It's important that the general concepts work, and I went for a broad range of those which might have made it too well rounded. Feedback rather welcome, especially if it's "tone it down you maniac, this is a wizard-class deck!".

The misplays are linked to spirits, as most of them have triggers when you play spirits and most of those triggers either make more sense if triggered before combat rather than after. Very often they make NO sense if triggered after combat, but this isn't the case here. Here it's just rather bad play that can cause the AI not to attack when it could or waste turns it had no cause to waste. I'm terrified of breaking anything, but when I apply tweaks for the current pack I'll certainly make all the spirits involved in the AI decks playable before and after combat (or in case the AI proves stubborn about misplays only before it depending on some factors).

Mind Control | Open

In order to test Priestess and Mind Stealer properly, I'll take mind control off. I've already taken invisibility off Conjurer and Troll Shaman for the same purpose.

A few words on Mind Control, besides that.

I can understand the appeal of it, and have had my ass kicked by it plenty of times, as when your deck scales it becomes the only thing that can really challenge you. So I can see how folks can grow attached to it.

However, I'm willing to bet good money that is not why it's in there, but rather a cool side effect. And the reason it was there, in the form of "Enemy X has mind control and it triggers rather often" doesn't exist anymore.

See, priestess had a wall deck, which would drag out games forever but it was very easy to prevent it from winning. It had to draw Animate Wall, and then cast it on the right thing, and then it would have to stick for the deck to really be able to get offensive. It was also uber-defensive, with Wall of Swords having both flying and stats that block even huge bombs and stuff like that, so the games against it were long, drawn out and unfun by design. They wouldn't give a tier2 enemy an ability which can murder you even when you're ready for wizards, they gave it to her because her deck was cool but painfuly passive and counterable.

Mind Stealer also had a deck which looked nasty on paper but was very easy to play against for many decks. Or it was considered too difficult to play against with creature decks. What the mind control does is let you not have to play against the enemy which has it. When you have a killer deck, this looks like a challenge.

The whole "mind control murders you" happens because the AI is generally bad at playing it's own decks in OG shandalar and the player is likely to stick everything simple and straightforward in it's deck (strongest cards in the OG pool were like that), so it was always an upside for the AI.

The problem I've encountered, though, is that now that the pool is huge, the tribute bug is gone and enemies have functional decks and beat you up all the time, what the AI gets with Mind Control is mostly terrible for it. If you have cards or combinations it can't play well, and there's heaps of those, you're playing vs. a seriously disadvantaged AI. No need for the Priestess to avoid it's own deck anymore, and Mind Stealer can win in a variety of ways against whatever you bring on with it's own deck.

So, despite the fact that it's only for testing purposes, this ability being tacked onto exclusively opponent X is pointless and counterproductive. I wouldn't mind seeing it around, having it crop up here and there at random, for example vs opponents you have a good score against, beseting you in dungeons as a debuff where you can't prep your deck with the foreknowledge it's gonna come up and things like that - the whole "now I have the moxes bwhahahahhha" incidental aspect is a very cool thing! But in practice, what the AI is gonna end up with is almost always going to be something it's going to be worse for it than it's own deck.

The reason I mention dungeons in this context is that the best place to have the ability around is one where it won't result in you winning cards / getting a duplicate card of your choice. Getting a taste of your own medicine is pretty cool, but the context of where and how this is appropriate has changed. Having it as a mandatory "trap" on a lair would probably be the best deal for randomly getting bushwacked by yourself while you go around looking for free stuff XD

Anyway, when I put up the whole deck pack, I'll put up the exe with the spirit tweaks (and other such tweaks) and invisibility and it removed for testing purposes. This is just something that's good to know. Also, the spirit tweaks are for testing purposes, too - they'll solve the spirit issues, for sure, but do they break something else in turn?

I'll do another one now before I double back on the Cleric again.

Right, here's a Prismat I'm happy with, still the same old dude, just with a seeminly more functional deck... and a new mana base.

Prismat Tweaks - new cheeze and new landbase | Open

I put in quite a bit of work into this thing, although it's hard to tell. Lots of stuff I tried didn't work, but some did and it seemed to fix issues I had in my notes.

New Land Base:

.164 2 Mountain
.188 2 Plains
.91 2 Forest

.1498 4 Temple Garden
.1710 4 Stomping Ground
.1708 4 Sacred Foundry

.5334 1 Battlefield Forge
.1882 1 Brushland
.1994 1 Karplusan Forest

.1203 4 Jungle Shrine

This is a bit ham fisted and needs proper refinement, and I did try out how the AI handles Forbidden Orchard, but this just made him extra trigger happy with it's X nukes and the benefits were marginal. This is certainly overkill, but it gets the job done.

The caveat is that he's somewhat easy-going with his mana base requirements as he usually only casts one spell per turn, a large amount of mana he uses over the course of the match is colorless and he's got no activation costs. Also, it's hard for him to screw his curve up as his response to pretty much anything is "sweep the board" anyway. And he also has lifegain. And ramp. He can afford to do nothing for a few turns. Piling up 12 dual lands and 4 Jungle Shrine seemed kinda adequate in Infinite Duel mode testing. So I'm not sure all this is necessary for him in particular and how well his potential success translates to guys who can't afford to play 16 potential taplands.

.116 4 Hurricane
.75 4 Earthquake
.337 4 Inferno

Those are as is. I'm not actually happy with their performance, being stuck with the wrong one in hand or facing mixed threats is a problem. The fact that he's very trigger happy with them is an issue too. Which is why he now has:

.290 3 Wrath of God - just so he doesn't blow cheap X based nukes early on all the time.
.5086 2 Restock - and so that he can get them back when he does, occasionally. Come to think of it, he could use more of these. Red card draw is missing from the pool, white card draw doesn't exist, and green card draw doesn't mix with his style, so this is his only option. It's a good option, but I'm not sure about his choice of stuff to return.

.4766 3 Skyshroud Claim
.318 3 Fellwar Stone

Those are necessary. The claim fetches shocklands, and the stone accelerates. His nukes are way too slow otherwise and it also helps with all the taplands.

.11833 3 Personal Sanctuary - This needs no explanation I suppose.

.12431 2 Natural End
.14284 4 Highland Game
.9355 3 Feudkiller's Verdict

And this is the lifegain. He wasn't using Dawnglow Infusion (and I think it's bugged but I'm not sure - maybe he was just having problems with paying for it properly, I had a note saying it needs replacing). Highland Game is a cheaper Onulet and doesn't have to be an artifact now that Living Artifact isn't in.

You might think that Feudkiller's Verdict is overkill, but I'm not sure. The deck has a big conceptual problem because of the AI. It doesn't have any proper card draw so it's a combo/control deck which is stuck with what it draws initially and then goes topdeck mode. But it doesn't have many sources of repeatable damage, and it's trigger happy with his one-shot sources. This just makes it empty the hand fast and sit there taking damage from anything the opponent has left in hand. Also - if facing fliers and it draws earthquake - problem, if it's the other way around - problems, etc etc. If problems - it loses life and can't use it's X spells right (unless Personal Sanctuary is down). So it needs something like, if not exactly, Feudkiller's Verdict , a jumbo Onulet (in the best case) to jumpstart it back into games that start off badly.

Also, in order to work it needs all the support cards, but this then means it can start the game with nothing but land search, lands and mana stones. If Burning Wish was implemented like it works IRL (meaning you can set it's 15 card sideboard personally), he'd have a Wrath, an Earthquake, a large card draw spell, a large lifegain spell etc. and 4 Burning Wish. In fact he'd only need 4 cards to choose from, and still run them properly. I do hope that works eventually, you can go places with those. The current implementation looked completely useless when I read it. Sorry if it took a lot of work, and ofc I understand it's the only thing you could do Korath! But unless I misunderstood it completely, what you implemented aren't wishes. Wishes tutor for specific cards you specifically picked for a specific deck with a specific purpose. But I totally get it if it's about how sideboarding works not being clear yet.

Feedback appreciated!

Ok, sun's up, I'm off to bed, I'll see about the Cleric Tomorrow. He might be fine as he is, so here's all 3. Plug them in and yell if they're too awful / too strong / something's wrong.

Well, took a bunch of time (going slower than I had hoped), but was overall a pleasant thing on the whole as I was doing things that have a wide selection of options - if one thing doesn't work, another might, just a matter of hitting the right stuff. I went through the rest of the dragons and also put in my first, clumsy attempt at a Dragonlord

First off the ones I got to say a few words about:

Whim | Open

What I did with Whim is keep him mostly the way he was, just tightened it's plan. It, sadly, no longer has a Psychic Venom equivalent, simply because the whole Royal Assassin / Meekstone / Icy Manipulator / Kismet engine was large enough as it was and fitting in Psychic Venom and, say, at least Pestermite was getting in the way of fitting in general deck smoothers. I did put in something very promising that fits in with the whole plan and keeps him interesting (and functions a lot like Psychic Venom)

.1206 4 Arcane Sanctum
.1713 4 Hallowed Fountain
.1709 4 Watery Grave
.1711 4 Godless Shrine
.126 2 Island
.188 2 Plains
.239 2 Swamp

That's the general, somewhat overkill mana base. What I added to it here was the three Artifact lands:

.1219 1 Ancient Den
.1216 1 Seat of the Synod
.1215 1 Vault of Whispers

Which can, ofc, be fetched with:

.7143 3 Trinket Mage

Who also, and this is rather important, fetches:

.159 3 Meekstone

Which plays nicely with:

.12895 4 Blind Obedience - which replaces BOTH Kismet and Psychic Venom, being a Kismet that also pings people whenever the deck plays something, and with all the redundant Blind Obedience, Meekstone , Fellwar Stone and such there's plenty of opportunity for just that. It impossibly perfect for this deck!

Not to mention that both of those play ball with:

.209 4 Royal Assassin - Main murderer besides Meekstone
.119 3 Icy Manipulator - Helps both of those out.

And all this is backed up by:

.318 3 Fellwar Stone - To help with playing stuff and paying for pings.
.1549 3 Compulsive Research - To draw cards and dump lands which become unneeded past some point.

As you can see, that's quite an engine and requires the deck to be as tight as it humanley can. To help out further I added the extremely value:

.7936 3 Ribbons of Night - which kills things, gains life, draws cards, washes dishes, walks the dog... and is currently bugged so the AI can throw it at the enemy head So I suppose I should take it out, but there's nothing really like it. I suppose I could stick Psychic Venom in it's place until the next update, will decide before I upload later today.

The creatures, which also play ball with the rest of the stuff are

.1362 3 Tidehollow Strix - Because it's an annoying fiend from hell.
.8663 3 Serra Avenger - Because it's a cheap vigilant flier, in a Meekstone deck, what's not to like.
.1364 3 Serra Sphinx - Because I'm so tired of Serra Angels and this guy is the blue dragon. Shame about Keiga not playing nice with the whole tapdown thing.

Seems to work quite well. I stayed away from Pestermite because all faeries are being misplayed at the moment, and if Pestermite is AI-risky then forcing Psychic Venom / Contaminated Ground shennanigans isn't nearly as rewarding as it would otherwise be, even with Icy Manipulator.

I LIKE those shennanigans, don't get me wrong, and anything that make Pestermite even more annoying than it is is a reason to get up in the morning... but the deck was ambitious as it was and I want to see if tightening it properly makes it breathe better.

Should be lots of f... er, shoud be bloody murderous

Then I took a look at Mandurang,

Mandurang | Open

and what I saw displeased me - the AI was likely to use Honden of Infinite Rage activations on creatures it couldn't kill with them, and blowing Smoldering Tar too easily. This isn't necessarily a hopeless situation, but I gave it a mana base overhaul to the new model, cut some weight and added 3 Kokusho, the Evening Star. Prismat can't play the green guy, Whim can't play the blue guy, Dracur doesn't really want to play the red guy and Kiska-Ra doesn't want to play the white guy. But at least Mandurang can play Kokusho like a champ. My talk about dragons below doesn't apply to some non-red dragons, and the ones from Kamigawa in particular - they're both good cards, diverse and interesting mechanically and exotic enough to override my natural reaction to dragons.

No reason to go full detailed, it was just a mana-base and light tweaks for now, he'll get a better looking at at a later date. It's still a rather functional and capable control deck, if the new mana base holds together.

Then I looked at Kiska-Ra.

Kiska-Ra | Open

And the notes (and a bunch of looking at it) made me realize that reworking him properly (or rather, just tweaking him right) was going to take more work than he really needs right now, as a bunch of other guys have more pressing problems.

I replaced Tawnos' Weaponry with Leonin Scimitar (friendlier casting and attach cost), found room for 4 Noble Hierarch and applied the new mana base tweaks.

The problem with the deck, which I pretty much just trimmed down to 60 cards originally, is that it's built to accomodate Pandora's Box too much and the Box doesn't pop up all that often (and can still screw it up). So I'd much rather shut the box off, give him a different starter, and rebuild accordingly.

Also, the other problem is that he's using Ancestral Recall and Timetwister to draw cards, and he only has one of each. It's hard to tell how good a deck is when it's like that, it wants more card draw and such card draw that doesn't swing games as much as those two things. So I'd have to account for that in the rework, too.

So for now, I just did what I did and left him like that.

And then I took a look at Dracur,

Dracur | Open

as I wasn't happy with him. He was neither bad or non-functional, but the purpose of remaking him was to make him less one-dimensional and griefy, and I remade him into something that was sorta-one dimensinal and griefy (And I learned a lot about the AI behavior since then so some things were popping up which I couldn't account for before).

This took a lot of time and builds until I got here.

.1710 4 Stomping Ground
.1707 4 Overgrown Tomb
.1714 4 Blood Crypt
.1204 4 Savage Lands
.91 2 Forest
.164 3 Mountain
.239 3 Swamp

Mana base's the new one, that part's fine.

.10755 4 Dragonmaster Outcast - I gave him 4 of these. He mishandles them ( baffling, I know), in a way that I can't really get worked up about. Long story short, he lets it die too easily. But I worked around that, I hope.

What the deck otherwise does is deal with your creatures:

.1746 4 Arc Lightning - For the little guys (gets thrown at the player, grrr)
.14590 4 Merciless Executioner - For the big guys
.5232 4 Flametongue Kavu - For the mid guys

And it tries to get those CiP kill something buggers, as well as the Dragonspeaker Shaman, back into play with:

.1223 3 Eternal Witness
.3546 3 Recurring Nightmare - which is overkill if used right, ofc.

Now, since the Executioner and Reccuring Nightmare need their sac fodder, and it also helps Flametongue Kavu enter play if nothing is on the board, I tried a bunch of stuff until I settled on:

.10078 3 Sprouting Thrinax
.11553 4 Viridian Emissary

And the Emissary, in theory and also practice, helps ramp out lands to get the Dragonspeaker Shaman active earlier, along with:

.4766 4 Skyshroud Claim - which fetches Shocklands.

With the ramp, the deck looked like it could use something splashy and way above the curve (apart from recasting Reccuring Nightmare) so I gave it:

.10096 3 Violent Ultimatum

To help deal with pesky enchantments, artifacts, big guys and also to call back to it's land destruction roots. It can also, in theory, recurr it with Eternal Withness.

So, there's a lot of stuff going on, it's got overkill coming out of it's ears, and will certainly need tweaking in the future, but it's a more interesting deck than it used to be. Also, there's a long list of stuff I tried out that can be put back in and the deck rearranged.

I hope it's enjoyable, I'll give it a few more hours before I put all these up, and I'll check in case something slipped my attention. I'm slightly worried that 4 executioner 4 kavu 4 recurring nightmare might be too many dead cards vs. some decks. Dragons can be a lot of things, but grossly uneven isn't one of them.

And, well, I only put in a Sainted one out of all the wizards (and, ofc, it's not guaranteed to work), but here's a very inept attempt at a Dragonlord. Mono red, ofc, all my wizards are gonna be mono.

Fist a personal note on why I'm not exactly qualified to make this deck:

Me and Dragons | Open

As a person I'm extremely indifferent to dragons. When I was a really young kid, just learned to read, I got crazy about dionsaurs. Knew all their names and all sorts of stuff about them before I knew much of anything else in the world. So when I ran into dragons I wondered if that was some kind of dinosaur I never learned about. Adults told me dragons were what folks who didn't know about dinosaurs thought the dinosaur bones belonged to, so they made them up. This made me, for the rest of my life, always kind of register dragons not as something fantastical or imaginative as much as something people were being smartass about with no qualifications some times in the past. I don't even think about it, and I sure dig a bunch of outlandish stuff from all sorts of mythologies, but my brain just never filed them down in "fantasy" but rather in "nonsense". Which is a completely random thing and I'm obviously in a minority that way.

And it makes me sorta unqualified to make proper dragon-themed decks. Dragons were mostly designed along the same way as deamons, red junk rares and curve toppers (sometimes having a block mechanic), and I only ever played them in limited (and the Shivan Dragon in Shandalar), never in constructed. They always kinda made them simple, because they were supposed to appeal to kids and if I was playing red I wasn't even looking for a 6+ mana flying blunt instrument. They don't actually make any sense in that color mechanically, and they never resonated with me aestethically or conceptually, so I had to go google and research various lists and interviews and all that to get an idea about which ones are worthwhile. I did know some and some were obviously good, but I couldn't tell what most of them were really for like I can with many other cards. Dragons are a complete blind spot for me, they kind of just happened to other people.

Also, the other problem is that they are, by design, non-competitive. A dragon-tribal deck is a silly idea (much like a demon-tribal deck would be), because you're wasting deck space on things you should only need several of (or even none at all) to win a match. I can easily imagine a teenager playing with his kid brother, the former with an all-deamon deck and the later with an all-dragon deck, doing nothing for the first 6 or so turns and then playing all these huge things trying to outdo each other who's guys are cooler... but trying to make the end-boss for red out of that? In a format with moxes prety much guaranteed for the player? Tall order.

Well, if we were talking mechanics and history, the end boss for red would be a goblin, but that can't be so it won't be. Even if they don't register with me, even if they're grossly limited in design and even if a dragon-themed deck is an idea destined for very uneven games depending on the draw (making it completely unsuitable to the guy you only ever get to fight once if he gets a bad hand and loses) I felt like it has to be dragons.

I mean, they made, what, 50 of those things? When something is that limited in usefulness and application and there's so much of it, you gotta suppose there's a huge audience for it that just wants to see a lot of flying reptiles breathing fire for some reason. I just don't have enough experience with decks that make so little sense mechanically, and especially with making them work at the level this one is supposed to work and also through a proxy (that being the AI), and as far as "nonsense" goes, a dragon deck is probably way up there. (And I'm not talking about dragonstorm and such, that's a whole different story).

Still, went and made one. Wasn't even annoyed, it's something I never ever done. I'm curious if I can find a way to make it work properly in mono red, too. Never too old to learn. Hope it makes folks happy, too. Or if it doesn't, I don't mind working on it until it does

Dragonlord | Open

.11264 4 Glimmerpost
.6637 4 Cloudpost

I went with these. The AI is reluctant to play them, like it's reluctant to play most colorless lands, but the deck being the way it is this turned out to be less of a problem in testing. Time will tell.

.10700 1 Valakut, the Molten Pinnacle
.164 14 Mountain

Also went with that. Might seem strange to put Valakut in with all the locuses, but I wasn't going for Mana Flare (which is plan b in case this crashes and burns), and the biggest reason for the OG dragonlord being the pile of crap it was was because of protection from red. If the game drags, Valakut saves.

.17 1 Black Lotus
.168 1 Mox Ruby
.230 1 Sol Ring

Those things went in. They're quite needed. Lotus maybe not so much, honestly, but can't hurt to try.

.510 1 Mishra's Workshop
.3962 4 Worn Powerstone

These two are also needed. Again, this is plan A, if it works - cool, if not - Mana Flare.

.4041 3 Quicksilver Amulet

And that is my (and probably anyone else's) other thing. Either the deck gets a Worn Powerstone and ramps stuff out with it, or it gets this thing out and starts plopping down dragons.

The Locuses, however, are for:

.1492 2 Dragon Roost

Which is what I'd personally make the deck about. Mana Flare, that thing, as many dragons as you want and the rest of the deck can be actual useful things. Dragonspeaker Shaman isn't in because he only kicks in when there are 6 lands down, and he wants to be in a deck with actual land ramp, not a deck with many high costed things. In a dragon-tribal deck he's a useless 1/1 that doesn't help get the useless expensive guys in there any faster.

The lads:

.6512 2 Dragon Mage - I... needed someting, anything, that draws cards somehow. If this guy is on the board, that means there's a lot of mana or a Quicksilver amulet down and the deck just wants to refill the hand and churn them out.
.7363 3 Ryusei, the Falling Star - This guy is useful in multiples, becuase one gets played as a 5/5 and the next one is an Earthquake for X = 5. So it's a way to have dragons in a deck but have at least one of them not eat deck space.
.1539 2 Bogardan Hellkite - Well, he's an actually good, if overcosted card. Since the deck can sneak him in cheaply, he's in.
.12629 3 Thundermaw Hellkite - He's sorta okish. Getting him out with the Quicksilver Amulet isn't much different than hard-casting him so he'd make a better curve-topper than a dragon in a pile of dragons, but he's there so there's something to reliably play without shennanigans.
.13563 3 Stormbreath Dragon - And he's there as both a mana sink and something you can't just block up with pro:red white dudes.

I suppose I could have squeezed in the viashino legend from mirage who recruits dragons for one-turn only, too. There's things that can go wrong with him, though.

.282 1 Wheel of Fortune - Quite welcome in this deck.

.14520 3 Bathe in Dragonfire - Because dragons, lol. Also it only hits creatures, which is AI friendly.
.13640 3 Sudden Demise - Because lots of mana around. Interested in how it works out.

.5433 2 Smash - Kills your mox and draws a card, what's not to like. Also:
.11530 2 Spine of Ish Sah - Kills your protection from red, and can be smashed to draw a card so smash isn't a dead card. Honestly, if this deck wasn't so loaded down with dragons, it would have 4 of these, and be much better for it as this thing (if it doesn't choke up the deck) can solve issues with wizards being mono-colored (provided the wizard has issues with that, and this one does).

So, I'm curious as to how it works out. My first ever DRAGONZ!!1!!2 deck.

Quick, anyone who has actual experience with them and mono-red decks like these - what did I mess up? Golden opportunity to call me names, I won't mind, I'd be glad to learn something new!

I'll do some more testing on them and put them up. Sorry it's taking so long.

---

Ok, so I did a bit more testing on these. Kiska-Ra is still the same old Kiska-Ra, Mandurang is basically the same and I left the bugged Ribbons of Night at Whim*. If any of them turns out to be easy to slap around, be sure to tell me. The cleric went with the cantrip plan, not the best cantrips ever, but it lets him cycle through the deck faster and occasionally make a cool play.

Dracur is a mixed bag. He's way too eager to sacrifice/chump block with the very important shaman. I've also seen him occasionally lock up with Reccurring Nightmare (there's a much weaker alternative, which might be a very good idea in general, but I'm curious what happens). I think I can get the deck to work if it happens not to, but I've also had it win fair and square.

Dragonlord is just silly. I'm a dragon noob and making a dragon deck is a silly idea in the first place. The good hands are good, but the bad hands can just have him sitting there doing nothing for a while. I think that I could put togather a good Mana Flare / Gauntlet of Power deck for it, though, along the lines of the SotA one, with Dragon Roost and fewer dragons. Also, some combination of that "all creatures lose flying" and Form of the Dragon would be cheezy as hell, but then it would just eat a fireball to the face. Too many bases to cover any way you look at it.

IDK I'd love to try this one out in practice, just to see how lousy I did, just save a game before the red castle and give it a few shots. Hard to tweak a wizard deck and hard to test it properly, too.

Here's the batch with the rest of the dragons, dragonlord and the cleric, enjoy. I think they're all better than they were in the previous batch. I'll do djinns for mana bases as a larger group (Queltosh and Aga Ganeer are kinda fine, Alt-A-Kesh was a mite too brutal and Saltem Tor was bit of a mess) along with Warlock and Nether Fiend.

*Whim also kinda clued me in on part of what was up with Centaur Chieftain - AI is kinda overeager to use the extort on Blind Obedience. This is fine in Whim, not that great at CW.

Sigh...

Spent the last couple of days working on djinns. The AI can't be trusted with anything other than alpha duals. The screwups go as far as to play Breeding Pool tapped first turn instead of paying 2 life and playing Birds of Paradise.

I'll have to backtrack through everything and change all the manabases back to what they were. What's worse, I'm not even sure this would make the AI good enough at handling a basic RUG deck. I tried more than I think I tried with any other deck so far from the start of all this - nothing. Pure awfulness. Enough stuff is there that something should've worked, but nothing works right, even stuff that works kinda right elsewhere. I didn't even have the strenght to do Queltosh, I just changed the mana base, ran 5 matches and saw that it's hopeless. Now I'm thikning that the cip-tapped trilands which I put in there the last time around were mistakes all along, too. It's impossible to untangle the mess, you nail one card which is causing cascades, replace it with something else, then something else causes cascades, and if nothing is too horrible - the deck doesn't work as a deck. FML.

And I got to install both Forge and Shandalar on a stronger machine. Moving from where the thinking time was to 4 times as much (half of the original) improved things visibly the first time around, the awfulness is owerwhelming enough that even a slim chance of improvement is worth it.

I've got to find where "how AI values" whatever it values is, to see it with my own eyes. And then fiddle around with it. Must've been some really epic compromises to result in some of the stuff I've seen, just in the last 2 days, I just have to see it to believe it. If some of this stuff could be changed without causing a zombie apocalypse it'd be worth it.

Took me a while to get a functional install onto another machine. I really got lucky the last time around, this time around I hit quite a few snags. The proper procedure, in broad terms, seems to be this:

1) Get Cirith Ungol's install.
2) Run his patcher.
3) Get windows c related files from one of the updates.
4) Get the 3 art updates from the same post where you got the install, apply them.
5) Get the shandalar update from Cirith Ungol's folder
6) Get current shandalar update

and then, ofc, put all my stuff back in.

And this would still leave someone without magic_updater, and possibly other things as well.

Whew. At least I tracked down the display bug to Cirith Ungols install, it's there even if you don't even do anything with Korath's stuff.

Also, something not everyone'll know is that you can make your own shortcut with comand line parameters. Silly and obvious, didn't occur to me before. This means you can make a "X:\...\Program\Shandalar --deckbuilder" shortcut which doesn't go away when you clean your memory ^^ I bet a million people go "doooooh" at this, yet a bunch find this a nice bit of info.

This means that, in theory, I could make a shortcut for testing any particular shandalar deck with one click.

---

Here's a tricky question, though. I'm using notepad++ to edit the decks. That is, I brainstorm/edit decks in Forge (with my custom shandalar set). Then I just pop open the deckbuilder and either edit an old deck or put it together quickly while alt tabbing.

But then I have to open the manalink-format deck, find the appropriate XXXX.deck in decks, open that, copy paste into it. Once they're open this is nice, becaue notepad++ keeps them open and updates any further edits from testing. I even put the relevant number in the "maker" tab for each deck so I see it when I open it first.

However, I remembered that I culd just put metadata on actual files, so that they display the name of the monster (or even, say, Tier 1 - Witch) in the explorer. This would be handy, but if I edit anything only the text gets saved and not the metadata. Notepad++ doesn't keep that.

Anyone know of a way to get around that? Would be a useful trick to know in general.

Even more useful would be to straight up be able to save decks in the shandalar format from the deckbuilder.

---

As for the curent round of tweaks - I redid the witch a bit. Not much, but I'm still trying to keep the pack free of .csv interventions. Not sure why, really. I also re-buffed Undead Knight, I kinda hit him too hard the last time around. Those two are kinda done, but I'm hesitant to upload anything right now before I got a big bunch of it. To avoid any situations like the one I might be in now, where I have to end up reverting all the mana base stuff back to duals.

I'm letting the Djinns stew a bit, too, they wore me out. The common, chronic, deep, general gameplay errors tend to affect red and green the most as those colors are really not well designed for how Shandalar AI does things. I've bumped the thinking time to 3X of what it was on the old machine, so I gotta see what happens, but in general green can't do it's core thing properly at all (if fight cards make it in and thus it gets removal then it might fare a bit better), and red cards are mostly not designed to be played the way AI plays them. That's partly why Aga Galneer, W/G/B works like a charm, while Saltem Tor U/G/R doesn't with objectively much stronger cards. Or why black decks in my pack tend to seem stronger than other ones - wide variety of effects in black by design, so the AI has acess to ones that it can play somewhat properly. Green only really works if you give it cards which are fundamentally not really green (see OG Great Druid or "how to put all the green fliers into one deck").

Gonna see about the Warlock and the Nether Fiend business.

Can you post all the link for the download so we can compare and update the working ones?

Also, be careful with Shandalar -d, deck will be created in the Playdeck folder so when testing duel deck you will either have to move your newly created deck in Decks folder or use additional pathing argument