Sets (MH2) Modern Horizons 2 Previews
- Thread starter TrainmasterGT
- Start date
CC down for maintenance. Might have to tag myself soon!
Onderzeeboot
Ecstatic Orb
What's 12 hours to a lifetime? "Soon" is all a matter of perspective (see also "Blizzard soon") 
You can always count on your pal Velrun! The tokens of Modern Horizons 2.
https://magic.wizards.com/en/articles/archive/card-preview/tokens-modern-horizons-2-2021-06-03
The tokens of Modern Horizons 2.https://magic.wizards.com/en/articles/archive/card-preview/tokens-modern-horizons-2-2021-06-03
What did they do to Dr. Disrespect?
Am I the only one refreshing CubeCobra every 2 minutes just to see whether it's up?
What do we think about Grief at Medium to Medium-High power levels? It feels borderline. I’ve tried Thoughtseize, but it was a little too efficient. I’ve almost run Unmask at times, but it always felt a little underwhelming. It’s way more interesting than running Duress or Inquisition of Kozilek which is my current solution. Grief is more abusable, but has a greater cost in the early turns. It‘s 4cmc and a 3/2 so it misses a chunk of recursion I run like:
The stars could align in an opening hand with Ephemerate, but I don’t run duplicates so I don’t expect that happens often. Flickering to hand via Shepherd of the Flock doesn’t hit the same.
Reanimating grief with a 4+cmc reanimation spell doesn’t seem particularly oppressive. I don’t run much instant speed bounce.
It feels like it might be fine?
The stars could align in an opening hand with Ephemerate, but I don’t run duplicates so I don’t expect that happens often. Flickering to hand via Shepherd of the Flock doesn’t hit the same.
Reanimating grief with a 4+cmc reanimation spell doesn’t seem particularly oppressive. I don’t run much instant speed bounce.
It feels like it might be fine?
It’s back!Am I the only one refreshing CubeCobra every 2 minutes just to see whether it's up?
only one way to find out for sure!
I think it's a little overbearing for what I want. I am leery of "free" spells in general. One recursion spell that it feels a little gross with is persist. The applicable blink suite widens quite a bit if you hard-cast it, or once you recur it the first time. You don't run some of the classic blink enablers like flickerwisp, but I do, and I also run Emiel the blessed. That turns grief into a hand lock, which.... isn't ideal. So it might be a little less abusable for you, and I would feel a bit safer testing it in the GYC cube.
For now I'm happy with the role that Duress and Collective Brutality play, and have been fairly happy with a low density of that type of effect. If I wanted a creature version of the effect for my cube I would probably turn to Kitesail Freebooter first
Onderzeeboot
Ecstatic Orb
Grief into Ephemerate is probably one of the filthiest openings you can do now in a fair cube.
i personally just don’t run hand disruption since there are no on the stack combos to police anyway (except control wincon Thassa’s Oracle but no turbo self mill to make it a fast combo). i think Grief is really good, but i’m not interested in ripping stuff out of hands on cube night. and yeah it has several nutty play lines that could come up in my environment if i did run it.
Still a couple random cards missing from MH2. I sent the devs an email.
Also asked when they plan on fixing draft bots lol. Seems like a fair set of lands for one (one!) UB deck to take.....
a grand total of 1 of these actually matters to their deck...
Yeah, the only repeatable blink I have is Soulherder which commits you to 3c with Grief. In general, my blink and reanimation package sounds very toned down compared to yours. Apprentice Necromancer and Priest of Fell Rites are the fastest way to reanimate Grief in my cube, and are easy to interact with.
I totally understand not wanting to run hand disruption (or LD). My cube has enough of a combo component to warrant the extra disruption. Many of these combo centerpieces are not creatures which black has a hard time dealing with. Without disruption, it becomes easy for games to become less interactive while each deck gets into a race to see who can spin their wheels fastest and execute their game plan. Since decks are less reliant on attacking in my cube than others, there needs to be multiple avenues to interact/disrupt your opponent's game plan.
I have also come to the realization that I prefer to disrupt via discard than to disrupt via graveyard hate. There are many ways to get discarded cards back in my cube, and there are a lot of times when having a card in the graveyard is almost as useful as having it in hand. This gives a little extra tension to discard. Even targeted discard can backfire...maybe the only nonland card someone has is an Honored Hydra or Deep Analysis or Golgari Grave-Troll, Squee, Goblin Nabob, etc. Not saying that alone makes Grief a fair card, but that along with my blink and reanimator suite being B/C-tiered has convinced me to test it. I think it should be pretty fair.
Also asked when they plan on fixing draft bots lol. Seems like a fair set of lands for one (one!) UB deck to take.....
a grand total of 1 of these actually matters to their deck...
I don't see why this continues to be a problem. It's so absurd.
I know!
Has anyone tested moving all dual lands into the multicolored sections? Maybe the draft algorithm would treat them like truly azorius cards for Hallowed Fountain, dimir for Watery Grave, etc.
If not I might try at some point. The draft bots are at least reasonably normal when it comes to picking multicolored cards. Not even sure that it works, but might be worth a try. If nobody's tried it, I might when I get a chance to do that much fiddling.
I think moving the lands helped! Still have some greedy ass bots, but they seem better at picking lands that actually benefit them (better lol, not 100%). Here's a Jund deck a bot made, and besides flooded grove, these lands are pretty good selections by and large:
And no lands in the SB that are worthless. Just a colorless manland. This URW deck did a little worse, but it was probably taking lands earlier when it was picking RG and other color cards:
This BGU deck had 100% success on selecting appropriate lands!:
So I think that moving the lands into the guild slots does help with drafting. A remaining issue is that the bots pick them crazy highly, so they can still end up with off-color lands based on one or two speculative picks. And I left fetchlands in the lands slots, because they aren't really dual-lands. Might have to move them though, just to get the bots to calm down on picking them a bit.
And no lands in the SB that are worthless. Just a colorless manland. This URW deck did a little worse, but it was probably taking lands earlier when it was picking RG and other color cards:
This BGU deck had 100% success on selecting appropriate lands!:
So I think that moving the lands into the guild slots does help with drafting. A remaining issue is that the bots pick them crazy highly, so they can still end up with off-color lands based on one or two speculative picks. And I left fetchlands in the lands slots, because they aren't really dual-lands. Might have to move them though, just to get the bots to calm down on picking them a bit.
Moved the fetches, and it did seem to help a bit more. For whatever reason 2 color bots are still pretty bad at picking lands, but three color decks seem to be doing at least a mediocre job of staying in their colors. I was seeing fetches pretty late in packs in my test draft (like later than pick 2 lol) so moving the lands seems to help, but only to a degree. I actually ended up with >1 fixing land in my deck, so seems like a positive change. Actually managed a decent three color mana base in the previous test draft.
Makes sense to me based off possible implementations they've done for draft AI in the back-end. They're not quite Skynet so any little bit to nudge them towards more specific decision trees when trying to analyze a given pick is sure to help. I'll probably be making the same adjustment later today.
I messaged the dev's about it. I think there mightve been a couple more missing, but don't remember specificallyI thought the whole set was up on CC, but I tried to add Necrogoyf and it isn't there
Talked to ruler501, she said the color identity of the lands is taken into account, but not color category, sort, or tags. I think you just got (un)lucky.
Bummer. Did she mention anything about fixing it? Because though she say it takes color identity into account, I see mono red bots with watery grave and similar.
I'm no coding expert by any stretch of the imagination, but it seems like these bots are too focused on drafting per what numerically looks right to draft, and not enough on trying to build a deck. Like a legit deck that a human would try to build, no offense to CCs... iffy deck builder. Like, isnt that what a drafter is trying to accomplish? They aren't trying to scope down every numerically best card in some sub category of the cube, they are trying to work together a deck piece by piece. That might mean passing the Daretti for a Shock, because you are in Rx aggro, not artifact reanimator.
I dunno. Just really wish they were better. It's basically the only data I can get right now, and it's not very good data.
FWIW, the bots look a lot better with this release to me. Most of the land picks are justifiable now (there weren't before), and they tend to be on color.
There is still something off when the pick would otherwise be mediocre: https://cubecobra.com/cube/deck/60ba5c42ebd0f4102849c258?seat=5&view=draftbot&pick=33
Here, Stomping Ground doesn't do absolutely anything, yet it's the highest rated pick (Sacred Foundry was picked because of randomness). The Usability Percent shouldn't be 100, or if it is set to 100 for lands, because they are technically "castable" by any deck, one of the other metrics should penalize an off-color land.
We start to get into "debatable" territory though. The bots ATM feel to me like a playgroup that really, really likes fixing, even when completely speculative, at the expense of worse 23rds.
There is still something off when the pick would otherwise be mediocre: https://cubecobra.com/cube/deck/60ba5c42ebd0f4102849c258?seat=5&view=draftbot&pick=33
Here, Stomping Ground doesn't do absolutely anything, yet it's the highest rated pick (Sacred Foundry was picked because of randomness). The Usability Percent shouldn't be 100, or if it is set to 100 for lands, because they are technically "castable" by any deck, one of the other metrics should penalize an off-color land.
We start to get into "debatable" territory though. The bots ATM feel to me like a playgroup that really, really likes fixing, even when completely speculative, at the expense of worse 23rds.
Yes, she's working on more improvements!
Thanks for the info! The usability percentage definitely seems like a big issue. Usability of lands should directly tie into the color spread of drafted cards; the "deck" the bot is working towards.
It's interesting that it seems to assign identical weights to lands as it does artifacts. I wouldn't expect that with color identity taken into account.
"Internal synergy", "openness" and "colors" also seem to be redundant, at least on this one pick. They all match within each other across the cards where one of them matches.
It's interesting that it seems to assign identical weights to lands as it does artifacts. I wouldn't expect that with color identity taken into account.
"Internal synergy", "openness" and "colors" also seem to be redundant, at least on this one pick. They all match within each other across the cards where one of them matches.