I've repeated myself on this topic a dozen times over, so let's just look at one simple scenario: you're planning to attend some competitive event in Modern. There is no meaningful or discernible way of determining what you might face. Why would you ever consider anything other than a fast proactive strategy? Unless you are participating in a small, or self-contained, predictable meta?
I would play the deck that I think I am best with. Other players appear to do the same. Just look at the evidence. We have lacked good MTGO data for two years now. In that time, we have seen people bring both fast and proactive decks, as well as slow and reactive decks, to major events. There was a long 2018 stretch where UW Control, about as slow and reactive as decks get, was the correct choice even with incomplete MTGO data. Now it's not, but we still don't have MTGO data. The presence or absence of data did not change. Nor did it make UW Control better or worse. Moreover, when we did have MTGO data in spades, players routinely made uninformed decisions about what to pilot at all Modern levels. An abundance of MTGO data did not suddenly lead to all players playing the best deck.
Overall, the presence or absence of data is not driving these deck decisions, and likely has little to no impact on metagame becoming fast and proactive. There is simply no demonstrable relationship between the amount of MTGO data and the majority of player deck decisions in Modern history. Most players just play whatever they want.
As I said earlier, I think you are unnecessarily grouping Modern criticisms. Modern might be too fast and linear. Wizards might need to release more data. But those have little, if any, connection with each other. You'd have a far stronger case arguing that Twin's absence makes the metagame too noninteractive than this unprovable allegation that a lack of MTGO data is causing it.
Why would you ever consider anything other than a fast proactive strategy? Unless you are participating in a small, or self-contained, predictable meta? Or you simply don't care about giving yourself the best odds, and just want to cross your fingers for good matchups?
The fast decks like Phoenix do some good at facing the unknown. Probably another reason why Burn decks have existed and thrived for years. I don't own a Phoenix deck, so just saying things from observation.
Don't understand much what you and Ktken are arguing about. However, it's a nice and cool Sunday afternoon. It's relaxing to post here while thinking what to do for tomorrow's New Year's eve celebrations. Just consider me a passerby reading posts here. :>
I would play the deck that I think I am best with.
Could agree with that. I play ponza, bgx, and a tier 3 uw control deck variant on sig - piloting them for awhile has made me got used to the tricks and lines of play to get out of sticky situations.... but give me something like Elves then you see a dumb beast playing.. always get 4 for 1'd when playing elves or pretty much any tribal deck. Feels better when I'm the one using the board wipes instead of being the one who has to dance around things like Anger of the Gods and Damnation.
You take KCI sure people might have board wipes and spot removal but how much artifact hate do they pack without white sideboard cards they probably can do nothing to stop you but oneshot your behind with creature combat thanks to a nuts draw.
I continue to try and build a system that can quantify interactivity based solely on a decklist. The most accurate system so far awards points based on interactive spells, defined as effects that target opposing cards or result in an opponent discarding, bouncing, destroying, sacrificing, exiling, etc. their own cards. For instance, 4 Bolts would earn a deck 4 points. 2 Verdicts would earn a deck 2 points. This system runs into a bunch of issues, most of which I am troubleshooting, but I want some input on two of the biggest:
1. The Burn problem
Burn is packed with cards that interact with opposing creatures and planeswalkers, but they generally are not used that way and the deck plays pretty non-interactively most of the time. How can we account for this in a scoring system? So far, the best correction I've come up with is a "preponderance of evidence" standard, i.e. that if a card is used more than 50% of the time in a non-interactive way, it wouldn't count. This means 4 Boros Charm does not award Burn 4 points, as it is most often used in a non-interactive fashion. Same with Lava Spike. But Bolt is flexible enough that we probably can't meet that Standard and would count it the full 4 points for Burn. This isn't a bad system but it's a bit subjective. Thoughts on adjustments?
2. The Tron problem
Tron has a surprising number of interactive cards. Just looking at GP ATL, this includes 4 Karns, 2 Ugins, 3 Ulamogs, 2 Ballistas, 1 Dismember, 4 O-Stones, and 1 Ghost Quarter for a total of 17 interaction points. By comparison, GDS at GP Portland had only 16 interactive points: 2 IoK, 4 TS, 1 Dismember, 3 Push, 2 Bolt, 3 Stubbs, 1 Terminate. Even if we include cantrips, we still can't make up the difference because Tron is packed with cantrip/search effects too. In fact, GDS and Tron basically have the same number of cantripping/dig effects. In my gut, I know that GDS interaction is different than Tron interaction, but I am struggling to quantify that difference. How would people quantify/operationalize it?
Curious to see what people think. I think this system has a lot of potential to get us out of the subjective weeds when it comes to defining interactivity, but it has some mechanical issues we need to figure out.
Imagine you are a non Standard player.
Outside of the US why would you buy into a top deck that was not just a bunch of multi-use staples if you did not have the cards already?
Once a year pptqs that could easily be sanctioned as sealed by your local LGS? A Modern grand prix every now and then in Europe which would cost more than the deck just to get to? The mkm series? That one random WMCQ that was modern? Actually that last one I went to the semis with in my small country.... with a landkill deck that preyed on all the efficient decks of the day like Affinity and Infect. Which rather proves the point, all those decks got wrecked by what appeared to them to be random.dec, there was no advantage in playing those acknowledged top decks if half the room was content to turn up with fringe decks, some of which had great matches against specific top decks. There is no incentive to invest into top decks, there is no real grind circuit available in the format outside of the US. The same is true of Legacy, with fewer opportunities for competitive play, but once you buy your big RL cards they maintain and ultimately gain value over the long term. In other words I can make money by owning key cards in the format, which makes me at least more likely to invest in them.
There is not much point building for a meta of top decks if people are slinging Soul Sisters et al, and most LGS stores in Europe at least are full of such mini metas with decks from years ago still about. It is what makes the idea of data analysis of the meta totally laughable outside of the US. There is no meta data that will be relevant to your local LGS, which is about the only place modern is occuring on this side of the pond. Until wotc give a reason to play Modern or a pathway for the format, things will continue this way, and I doubt if they want to do so.
Oh, and in response to measuring interactivity, I don't think you can quantify it. Interactivity is very, very subjective. You can not measure it any more than you can measure fun. People either feel they had an interactive game or they don't, but people experiencing the same game will view it differently. Cards that are interactive can be used very uninteractively.
It is very difficult to pinpoint when they are interactive abd when they are not.
There are things you can measure and things you can't, the big fallacy of so much politics is that everything can be meaningfully measured, and to a high degree of precision. Measuring win rates is a lot easier than interactivity, for the reasons highlighted in the post above.
I know it's probably a good chunk of work but for situations like Burn, would you consider awarding points based on the matchup? I think if burn is against something like Devoted Vizier, it's more likely to play Bolt as an interactive spell, so for this you would award 1 point per card. Against Lantern though, it's almost strictly to the face, so no points awarded. If Devoted Vizier is 75% of the meta and Lantern is 25% (obviously waaay simplifying), use the meta percentages to define it's general interactivity as a weight and multiply it by the number of bolts the deck uses. You could go a step further and instead of saying "Bolt is interactive against Vizier 51% of the time so it gets 1 point per Bolt vs Vizier" you can say "Bolt is interactive against Vizier 75% of the time and Vizier is 75% of the meta (holding the rest of the meta 100% uninteractive), so Bolt will receive .75(meta%)*.75(purpose%) + [other matchups calculated similarly]."
I honestly have no idea if what I just typed out makes sense, but unfortunately I think how burn plays out is very matchup dependent which makes it, in turn, very meta dependent. If there's a better way to account for what I'm trying to impart, I'd be interested in hearing it for my own math learning, but this is the best I can describe what I'm getting at. I don't think this is an instance where you can remove bias entirely, but I do think you can minimize it.
I don't think I can wrap my head around giving a detailed suggestion for Tron, but I think Tron feels different because people often scoop as if it were combo. While it makes sense to scoop to a ridiculous board state (say T3 Karn, T4 Ugin vs 2 lands), I think what you are actually scooping to is the interaction itself. With GDS you are not scooping to a single Stub or Kommand or even many of them, instead you can see they have a 12/12 and they have actually mounted an assault on your life total (rules that end the game). Tron doesn't always do that, so it's interaction is actually it's win condition. I would almost measure it similar to control in that sometimes people scoop T7 to control when they have no cards in hand against control with 5+. It's fairly well established that control can win with a Scornful Egotist once it has control, so realistically it's win condition can be construed to be it's interaction. I would make a similar claim for Tron in many cases, because while Tron is not winning quickly, it can force a scoop by creating a board state that makes their win condition irrelevant.*
I am curious how you are registering cards that interact multiple times, like Lavamancer or Valakut. The answer you're looking for may be between how you quantify multi-use removal and viewing part of Tron's interaction as some of it's win condition.
*I am very of the opinion that a win condition is a card that actually forces the game to end by game rules. This can be a 1/1 creature with a massive drawback, but it can never be Karn because Karn does not actually have an ability that forces the game to end by game rules. Additionally I generally rule out things like Teferi infinitely tucking himself so the opponent mills, which would also mean I do not consider Teferi a win condition because he is not the card that is actually ending the game; Lantern Control, however, has a real game plan to mill their opponent. I would 100% agree that Teferi "wins the game" but only by facilitation, not by execution; therefore an additional card is needed to secure the win. Just an odd difference in definition I've noticed when many people talk about win conditions. I only bring this up because I think it makes the Tron thought a bit more clear.
EDIT: While I've played a bit of Tron for testing, I'd love to hear someone's opinion on my thoughts who plays it as a primary deck.
Curious to see what people think. I think this system has a lot of potential to get us out of the subjective weeds when it comes to defining interactivity, but it has some mechanical issues we need to figure out.
How is it any less subjective than what Hoogland and others were criticized for doing? It's just "rigorous analysis" of made up, subjective numbers, right?
The Burn example really shows an irreconcilable problem, in that every card would need to be weighted with intent. That intent would change deck to deck and opponent to opponent. In Jeskai v Tron, Lightning Bolt goes to the face 99% of the time. In Burn v Infect, Lightning Bolt goes to a creature 99% of the time. Is Lightning Bolt an interactive card? It's a big "it depends."
There is no escaping the subjectivity of trying to guage interaction, because even we account for every single one of these altering attributes, it's still just analysis of a judgment call someone is making. At which point, the entire analysis is just a fancy numbers expression of opinion (exactly what Jeff Hoogland posted).
Curious to see what people think. I think this system has a lot of potential to get us out of the subjective weeds when it comes to defining interactivity, but it has some mechanical issues we need to figure out.
How is it any less subjective than what Hoogland and others were criticized for doing? It's just "rigorous analysis" of made up, subjective numbers, right?
The Burn example really shows an irreconcilable problem, in that every card would need to be weighted with intent. That intent would change deck to deck and opponent to opponent. In Jeskai v Tron, Lightning Bolt goes to the face 99% of the time. In Burn v Infect, Lightning Bolt goes to a creature 99% of the time. Is Lightning Bolt an interactive card? It's a big "it depends."
There is no escaping the subjectivity of trying to guage interaction, because even we account for every single one of these altering attributes, it's still just analysis of a judgment call someone is making. At which point, the entire analysis is just a fancy numbers expression of opinion (exactly what Jeff Hoogland posted).
I'd both agree and disagree. While I do think it's still very subjective, I didn't enjoy seeing what Hoogland put out.
This might be an interesting time to ask KTK to show his work and put out the numbers in say an excel spreadsheet that we can take and adjust the numbers in ourselves. So if you think Burn is less interactive than the number KTK put in you can go in and adjust it to what you think is right, then we can all see how much our opinion is swaying from one person to the next.
Counter argument to putting such a tool out is that we'll all likely suddenly think we're experts. I can dream though.
For sure interactivity questions, I think the easiest solution would be to add a third category. Instead of simply "interactive" and "non interactive" try "non interactive" "interactive threat" and "interactive non-threat" with each category worth a different amount of points (maybe 0, 1, 2 is correct, or 0, .75, 1.25, or whatever is better)
I would never consider lava spike an interactive card (its just going upstairs, its less interactive than any vanilla creature), but boros charm is sometimes interactive, so maybe thats enough to put it into the second category. Bolt is somewhere between threat/nonthreat depending on the deck as well.
This also somewhat solves the tron problem.
In theory, ugin is an interactive card, but in practice, its a coinflip of a few bolts/EEs or just straight up wins the game because the other player can't possibly beat it, so it should be worth less points than O stone, but more points than urza's tower.
Curious to see what people think. I think this system has a lot of potential to get us out of the subjective weeds when it comes to defining interactivity, but it has some mechanical issues we need to figure out.
How is it any less subjective than what Hoogland and others were criticized for doing? It's just "rigorous analysis" of made up, subjective numbers, right?
The Burn example really shows an irreconcilable problem, in that every card would need to be weighted with intent. That intent would change deck to deck and opponent to opponent. In Jeskai v Tron, Lightning Bolt goes to the face 99% of the time. In Burn v Infect, Lightning Bolt goes to a creature 99% of the time. Is Lightning Bolt an interactive card? It's a big "it depends."
There is no escaping the subjectivity of trying to guage interaction, because even we account for every single one of these altering attributes, it's still just analysis of a judgment call someone is making. At which point, the entire analysis is just a fancy numbers expression of opinion (exactly what Jeff Hoogland posted).
I shouldn't need to explain the difference, and I'm a little puzzled why I have to. Moreover, it's particularly unfounded and bizarre that you are challenging the entire proposed system without seeing any examples of it applied, and when I literally said it was still a work in progress.
1. Such a system would transparently define all terms and ratings and would be fully auditable by readers. Hoogland just grouped a bunch of stuff and didn't bother to explain the terms or groupings.
2. Such a system would be applied across multiple eras of Modern to test for false positives/negatives. Maybe even across formats. Hoogland just loooked at a handful of T8s and made a sweeping format generalisation without comparing his claim to other time frames if it's even a valid analysis method.
3. I am anticipating objections and flaws with the method. Hoogland just took it as gospel in revealing some big Modern reality without acknowledging any other competing interpretations.
These are just a few differences that should be obvious in just the post I wrote about this. Most rating systems start out as subjective, but if they actually measure what they claim, they become much more objective measures. ELO is an excellent example of this. If the proposed system does as promised, great, we can use it. If it makes too many false positives and negatives, we'll try to refine it or move on.
@Others
Thanks for the constructive feedback. I think those are interesting starting considerations. Matchup context is particularly interesting, but might be too complicated. More work is clearly needed!
So subjective analysis is fine. So long as you provide enough numbers for your subjective analysis.
As a math teacher who uses Excel spreadsheets for fun, the idea of doing this seems fascinating. But as a useful tool for anything other than an interesting thought experiment? Not as much. Have fun with it. I know I would.
So subjective analysis is fine. So long as you provide enough numbers for your subjective analysis.
As a math teacher who uses Excel spreadsheets for fun, the idea of doing this seems fascinating. But as a useful tool for anything other than an interesting thought experiment? Not as much. Have fun with it. I know I would.
So all rating/ranking scales are just subjective and useless thought experiments unless they rate/rank purely objectivr, ordinal numbers? We can't use any quantitative systems to measure or capture seemingly qualitative phenomena? This is the sort of reductionist logic which in a few sentences dismisses the foundation of most sports, social science, political, and economic scales we use. Even predictive web search ranking algorithms seem to just be useless thought experiments given your attitude. As such, I don't find it particularly credible. If a system is valid, reliable, transparent, replicatable, and meets other standards, it ceases to be "subjective" in the dismissive way you suggest and becomes a valuable analysis tool. Note that Hoogland's absurd pie chart with its extremely opaque categories does not attempt to meet this bar.
If I had posted the system with examples and it did not meet those benchmarks, I would understand your challenges. Given that I have not done that and you have already rejected it as not being a useful tool, I find it pretty easy to determine your objections are at best premature, baseless, and unhelpful.
EDIT: that MTGO analysis is a perfect example of a rigorous, data-driven approach to answering the question of format speed. Note that Hoogland's terrible analysis of 66% of decks trying to win/set up an insurmountable game state by T3 is largely rebutted by this graph, which shows that only 8% of games end on or before T3, and only 22% end on T4. Who cares what decks are trying to accomplish by some subjective standard when it cant even remotely predict the turn that games end on?
Regardless, I don't know who could look at either of these and think: "You know, I think I want to play a deck whose win condition is a planeswalker ultimate on turn 10, or a 4/4 that I can't activate until turn 6, but realistically don't activate it until 10+."
But who cares when decks are capable of winning? How is that meaningful in any way? It's even more meaningless when we have no clue how Hoogland defined those decks or even what decks are in that percent. It is significantly more valuable to know the actual turn distribution that games end, as this far more useful analysis shows.
The moment I'm home, I will see how this 2018 distribution compares to the distribution in 2015 Modern, which I know many view as the diverse golden age of multiple pillars with a low incidence of linear decks. That will allow us to see how Modern's speed has or has not changed since 2015 and the year of Twin, BGx, Bloom, and other 2015 mainstays.
But who cares when decks are capable of winning? How is that meaningful in any way? It's even more meaningless when we have no clue how Hoogland defined those decks or even what decks are in that percent. It is significantly more valuable to know the actual turn distribution that games end, as this far more useful analysis shows.
The moment I'm home, I will see how this 2018 distribution compares to the distribution in 2015 Modern, which I know many view as the diverse golden age of multiple pillars with a low incidence of linear decks. That will allow us to see how Modern's speed has or has not changed since 2015 and the year of Twin, BGx, Bloom, and other 2015 mainstays.
If you're looking to make a deck choice decision, are the two really any different in their conclusions? Since I edited my last post to include an extra line, I'll paste it here again, since it directly addresses the practical uses of data like this: I don't know who could look at either of these and think: "You know, I think I want to play a deck whose win condition is a planeswalker ultimate on turn 10, or a 4/4 that I can't activate until turn 6, but realistically don't activate it until 10+."
But who cares when decks are capable of winning? How is that meaningful in any way? It's even more meaningless when we have no clue how Hoogland defined those decks or even what decks are in that percent. It is significantly more valuable to know the actual turn distribution that games end, as this far more useful analysis shows.
The moment I'm home, I will see how this 2018 distribution compares to the distribution in 2015 Modern, which I know many view as the diverse golden age of multiple pillars with a low incidence of linear decks. That will allow us to see how Modern's speed has or has not changed since 2015 and the year of Twin, BGx, Bloom, and other 2015 mainstays.
If you're looking to make a deck choice decision, are the two really any different in their conclusions? Since I edited my last post to include an extra line, I'll paste it here again, since it directly addresses the practical uses of data like this: I don't know who could look at either of these and think: "You know, I think I want to play a deck whose win condition is a planeswalker ultimate on turn 10, or a 4/4 that I can't activate until turn 6, but realistically don't activate it until 10+."
They might say "Those other decks seem ill-equipped to win the long game, so if I can tune my answers to keep me alive until T5 I have a really good chance of leveraging stronger late-game win conditions." Both of those hypothetical and rhetorical positions are equally plausible.
Incidentally, as Kathal predicted, there is a very small difference in the turn distribution between 2015 and 2018 Modern. See the numbers below. An earlier version of this list combined the T0/T1/T2 numbers but I realized separating them is clearer. I'm not really sure what a T0 win is if neither player has a turn (seems like someone just forfeited before cards were played), so I might need to just remove that line entirely because it might not make sense.
2015 vs. 2018 cumulative win percentages by Turn N T0: 0% (2015) vs. 0.5% (2018) T1: 0% (2015) vs. 0.9% (2018) T2: 0.9% (2015) vs. 2.4% (2018) T3: 6.2% (2015) vs. 8.3% (2018) T4: 21.4% (2015) vs. 22.2% (2018) T5: 39.4% (2015) vs. 39.4% (2018) T6: 55.7% (2015) vs. 55.2% (2018) T7: 69.4% (2015) vs. 67.6% (2018) T8: 79% (2015) vs. 77% (2018) T9: 85.5% (2015) vs. 83.9% (2018) T10: 100% (2015) vs. 100% (2018)
The biggest points of difference are in the T0-T3 wins and the T9/T10+ wins. In 2018, 2.2% more games end on T0-T3 than in 2015, which indicates a slight increase in early game speed. That said, 2018 also sees 2% more games go beyond turn 9+, which means the long games go longer than they used to. This ultimately results in no difference between the average turn on which games end: using weighted averages, that would be turn 6.43 in 2015 and turn 6.42 in 2018.
I am really curious to see how this distribution compares to Standard, which I know he is also working on. I suspect Standard will be slower but only by 1-2 turns, at least if my MTG Arena experience is any indication (which it might not be because I predominantly play BO1 which has different dynamics).
2015 vs. 2018 cumulative win percentages by Turn N T0: 0% (2015) vs. 0.5% (2018) T1: 0% (2015) vs. 0.9% (2018) T2: 0.9% (2015) vs. 2.4% (2018) T3: 6.2% (2015) vs. 8.3% (2018) T4: 21.4% (2015) vs. 22.2% (2018) T5: 39.4% (2015) vs. 39.4% (2018) T6: 55.7% (2015) vs. 55.2% (2018) T7: 69.4% (2015) vs. 67.6% (2018) T8: 79% (2015) vs. 77% (2018) T9: 85.5% (2015) vs. 83.9% (2018) T10: 100% (2015) vs. 100% (2018)
I wish I had the patience to go and find all the posts and comments about how banning Twin would slow down the format and allow you to tap out on turn 3!
2015 vs. 2018 cumulative win percentages by Turn N T0: 0% (2015) vs. 0.5% (2018) T1: 0% (2015) vs. 0.9% (2018) T2: 0.9% (2015) vs. 2.4% (2018) T3: 6.2% (2015) vs. 8.3% (2018) T4: 21.4% (2015) vs. 22.2% (2018) T5: 39.4% (2015) vs. 39.4% (2018) T6: 55.7% (2015) vs. 55.2% (2018) T7: 69.4% (2015) vs. 67.6% (2018) T8: 79% (2015) vs. 77% (2018) T9: 85.5% (2015) vs. 83.9% (2018) T10: 100% (2015) vs. 100% (2018)
I wish I had the patience to go and find all the posts and comments about how banning Twin would slow down the format and allow you to tap out on turn 3!
Has anyone ever argued that banning Twin slowed down the format? I don't recall those arguments, and if I ever personally made that argument, I chalk it up to temporary insanity. Banning Twin obviously ran the risk of speeding up the format. I'm sure I've even acknowledged that in the past. We just didn't know how much it would speed up the format, if at all. But now we have numbers to assess how much faster the format has become following/as a result of that ban.
Tapping out on T3 does not impact T0, T1, or T2 wins (you can still tap out against Twin on T1/T2 with no fear of Twin winning the next turn). So that means we just need to look at the relative difference in T3 wins to see the win-turn difference in Twin and no-Twin Modern. In 2015, 5.3% of all games ended on T3. In 2018, it was 6%. Even if we can chalk all 100% of that .7% difference to Twin, which I doubt we can, that's still a pretty tiny margin. It's also not a very persuasive case to unban Twin. There are definitely good arguments to support an unban, but claiming that unbanning Twin will slow down Modern to the tune of .7% of games is not exactly compelling.
Has anyone ever argued that banning Twin slowed down the format?
Perhaps not directly, but at least one of the things dozens of people were saying how much more playable 4-drops would be as a result of Twin being gone, implying that larger cost spells on curve were more viable due to not dying quickly.
I remember this vividly because people have been saying all sorts of stupid and untrue things related to Twin for at least 3 years. And I've been there the whole time listening to it and dealing with it.
That turn 3 or earlier win percentage looks higher than I feel it should be in a 'turn 4' format for the overall format. I could understand if a few specific decks had wins that were around that level being on such early turns, but the format as a whole? Seems to me like the game isn't a turn 4 format if over 5% of wins happen before that in the format as a whole, even in the 2015 data set, and much worse being higher than 8% in the 2018 data set. Sure, most of the wins might be turn 5, but that is way more than enough wins to be a serious threat to win percentages for decks up against those fast decks, such that they have to treat turn 3 or earlier wins as a serious threat.
This plays into my earlier comments about counterspells, and how turn 2 (on the draw) ones play a different role to turn 3+ (on the draw) ones, and how Absorb will likely be competing for slots with Cryptic Command, rather than the 2 cmc stuff. I really hope WotC starts pushing some better 1-2 cmc main-board worthy answers into the format and push those turn 3 or earlier games down below 1% overall, and only around 5% for specific faster decks (which would win turn 4 on their faster games most of the time outside of nut draws and the opponent drawing terribly despite mulliganing as much as is worth the risk).
Only a few hours before New Year. Oh, we're back to Twin again. Although the discussion is more calm this time.
Just a little input on it. Not really debating anyone. When Twin got banned it's hard to say if the format got faster or more slow. Right after the ban people here in our area became more complacent tapping out on turn 4.. slower aggro decks without counterspells like Elves became more common. Well, decks like Affinity, Living End, and Burn still tried to win as fast as they can.. but you can see that they're no longer anxious tapping out turn 4 against anything with red and blue in the manabase. When twin was alive, those decks often have something like a galvanic blast, beast within, or combust ready to kill a Deceiver Exarch.
I would play the deck that I think I am best with. Other players appear to do the same. Just look at the evidence. We have lacked good MTGO data for two years now. In that time, we have seen people bring both fast and proactive decks, as well as slow and reactive decks, to major events. There was a long 2018 stretch where UW Control, about as slow and reactive as decks get, was the correct choice even with incomplete MTGO data. Now it's not, but we still don't have MTGO data. The presence or absence of data did not change. Nor did it make UW Control better or worse. Moreover, when we did have MTGO data in spades, players routinely made uninformed decisions about what to pilot at all Modern levels. An abundance of MTGO data did not suddenly lead to all players playing the best deck.
Overall, the presence or absence of data is not driving these deck decisions, and likely has little to no impact on metagame becoming fast and proactive. There is simply no demonstrable relationship between the amount of MTGO data and the majority of player deck decisions in Modern history. Most players just play whatever they want.
As I said earlier, I think you are unnecessarily grouping Modern criticisms. Modern might be too fast and linear. Wizards might need to release more data. But those have little, if any, connection with each other. You'd have a far stronger case arguing that Twin's absence makes the metagame too noninteractive than this unprovable allegation that a lack of MTGO data is causing it.
The fast decks like Phoenix do some good at facing the unknown. Probably another reason why Burn decks have existed and thrived for years. I don't own a Phoenix deck, so just saying things from observation.
Don't understand much what you and Ktken are arguing about. However, it's a nice and cool Sunday afternoon. It's relaxing to post here while thinking what to do for tomorrow's New Year's eve celebrations. Just consider me a passerby reading posts here. :>
Could agree with that. I play ponza, bgx, and a tier 3 uw control deck variant on sig - piloting them for awhile has made me got used to the tricks and lines of play to get out of sticky situations.... but give me something like Elves then you see a dumb beast playing.. always get 4 for 1'd when playing elves or pretty much any tribal deck. Feels better when I'm the one using the board wipes instead of being the one who has to dance around things like Anger of the Gods and Damnation.
Nexus MTG News // Nexus - Magic Art Gallery // MTG Dual Land Color Ratios Analyzer // MTG Card Drawing Odds Calculator
Want to play a UW control deck in modern, but don't have jace or snaps?
Please come visit us at the Emeria Titan control thread
I continue to try and build a system that can quantify interactivity based solely on a decklist. The most accurate system so far awards points based on interactive spells, defined as effects that target opposing cards or result in an opponent discarding, bouncing, destroying, sacrificing, exiling, etc. their own cards. For instance, 4 Bolts would earn a deck 4 points. 2 Verdicts would earn a deck 2 points. This system runs into a bunch of issues, most of which I am troubleshooting, but I want some input on two of the biggest:
1. The Burn problem
Burn is packed with cards that interact with opposing creatures and planeswalkers, but they generally are not used that way and the deck plays pretty non-interactively most of the time. How can we account for this in a scoring system? So far, the best correction I've come up with is a "preponderance of evidence" standard, i.e. that if a card is used more than 50% of the time in a non-interactive way, it wouldn't count. This means 4 Boros Charm does not award Burn 4 points, as it is most often used in a non-interactive fashion. Same with Lava Spike. But Bolt is flexible enough that we probably can't meet that Standard and would count it the full 4 points for Burn. This isn't a bad system but it's a bit subjective. Thoughts on adjustments?
2. The Tron problem
Tron has a surprising number of interactive cards. Just looking at GP ATL, this includes 4 Karns, 2 Ugins, 3 Ulamogs, 2 Ballistas, 1 Dismember, 4 O-Stones, and 1 Ghost Quarter for a total of 17 interaction points. By comparison, GDS at GP Portland had only 16 interactive points: 2 IoK, 4 TS, 1 Dismember, 3 Push, 2 Bolt, 3 Stubbs, 1 Terminate. Even if we include cantrips, we still can't make up the difference because Tron is packed with cantrip/search effects too. In fact, GDS and Tron basically have the same number of cantripping/dig effects. In my gut, I know that GDS interaction is different than Tron interaction, but I am struggling to quantify that difference. How would people quantify/operationalize it?
Curious to see what people think. I think this system has a lot of potential to get us out of the subjective weeds when it comes to defining interactivity, but it has some mechanical issues we need to figure out.
Outside of the US why would you buy into a top deck that was not just a bunch of multi-use staples if you did not have the cards already?
Once a year pptqs that could easily be sanctioned as sealed by your local LGS? A Modern grand prix every now and then in Europe which would cost more than the deck just to get to? The mkm series? That one random WMCQ that was modern? Actually that last one I went to the semis with in my small country.... with a landkill deck that preyed on all the efficient decks of the day like Affinity and Infect. Which rather proves the point, all those decks got wrecked by what appeared to them to be random.dec, there was no advantage in playing those acknowledged top decks if half the room was content to turn up with fringe decks, some of which had great matches against specific top decks. There is no incentive to invest into top decks, there is no real grind circuit available in the format outside of the US. The same is true of Legacy, with fewer opportunities for competitive play, but once you buy your big RL cards they maintain and ultimately gain value over the long term. In other words I can make money by owning key cards in the format, which makes me at least more likely to invest in them.
There is not much point building for a meta of top decks if people are slinging Soul Sisters et al, and most LGS stores in Europe at least are full of such mini metas with decks from years ago still about. It is what makes the idea of data analysis of the meta totally laughable outside of the US. There is no meta data that will be relevant to your local LGS, which is about the only place modern is occuring on this side of the pond. Until wotc give a reason to play Modern or a pathway for the format, things will continue this way, and I doubt if they want to do so.
Oh, and in response to measuring interactivity, I don't think you can quantify it. Interactivity is very, very subjective. You can not measure it any more than you can measure fun. People either feel they had an interactive game or they don't, but people experiencing the same game will view it differently. Cards that are interactive can be used very uninteractively.
It is very difficult to pinpoint when they are interactive abd when they are not.
There are things you can measure and things you can't, the big fallacy of so much politics is that everything can be meaningfully measured, and to a high degree of precision. Measuring win rates is a lot easier than interactivity, for the reasons highlighted in the post above.
I honestly have no idea if what I just typed out makes sense, but unfortunately I think how burn plays out is very matchup dependent which makes it, in turn, very meta dependent. If there's a better way to account for what I'm trying to impart, I'd be interested in hearing it for my own math learning, but this is the best I can describe what I'm getting at. I don't think this is an instance where you can remove bias entirely, but I do think you can minimize it.
I don't think I can wrap my head around giving a detailed suggestion for Tron, but I think Tron feels different because people often scoop as if it were combo. While it makes sense to scoop to a ridiculous board state (say T3 Karn, T4 Ugin vs 2 lands), I think what you are actually scooping to is the interaction itself. With GDS you are not scooping to a single Stub or Kommand or even many of them, instead you can see they have a 12/12 and they have actually mounted an assault on your life total (rules that end the game). Tron doesn't always do that, so it's interaction is actually it's win condition. I would almost measure it similar to control in that sometimes people scoop T7 to control when they have no cards in hand against control with 5+. It's fairly well established that control can win with a Scornful Egotist once it has control, so realistically it's win condition can be construed to be it's interaction. I would make a similar claim for Tron in many cases, because while Tron is not winning quickly, it can force a scoop by creating a board state that makes their win condition irrelevant.*
I am curious how you are registering cards that interact multiple times, like Lavamancer or Valakut. The answer you're looking for may be between how you quantify multi-use removal and viewing part of Tron's interaction as some of it's win condition.
*I am very of the opinion that a win condition is a card that actually forces the game to end by game rules. This can be a 1/1 creature with a massive drawback, but it can never be Karn because Karn does not actually have an ability that forces the game to end by game rules. Additionally I generally rule out things like Teferi infinitely tucking himself so the opponent mills, which would also mean I do not consider Teferi a win condition because he is not the card that is actually ending the game; Lantern Control, however, has a real game plan to mill their opponent. I would 100% agree that Teferi "wins the game" but only by facilitation, not by execution; therefore an additional card is needed to secure the win. Just an odd difference in definition I've noticed when many people talk about win conditions. I only bring this up because I think it makes the Tron thought a bit more clear.
EDIT: While I've played a bit of Tron for testing, I'd love to hear someone's opinion on my thoughts who plays it as a primary deck.
"Reveal a Dragon"
How is it any less subjective than what Hoogland and others were criticized for doing? It's just "rigorous analysis" of made up, subjective numbers, right?
The Burn example really shows an irreconcilable problem, in that every card would need to be weighted with intent. That intent would change deck to deck and opponent to opponent. In Jeskai v Tron, Lightning Bolt goes to the face 99% of the time. In Burn v Infect, Lightning Bolt goes to a creature 99% of the time. Is Lightning Bolt an interactive card? It's a big "it depends."
There is no escaping the subjectivity of trying to guage interaction, because even we account for every single one of these altering attributes, it's still just analysis of a judgment call someone is making. At which point, the entire analysis is just a fancy numbers expression of opinion (exactly what Jeff Hoogland posted).
UR ....... WUBR ........... WB ............. RGW ........ UBR ....... WUB .... BGU
Spells / Blink & Combo / Token Grind / Dino Tribal / Draw Cards / Zombies / Reanimate
This might be an interesting time to ask KTK to show his work and put out the numbers in say an excel spreadsheet that we can take and adjust the numbers in ourselves. So if you think Burn is less interactive than the number KTK put in you can go in and adjust it to what you think is right, then we can all see how much our opinion is swaying from one person to the next.
Counter argument to putting such a tool out is that we'll all likely suddenly think we're experts. I can dream though.
"Reveal a Dragon"
I would never consider lava spike an interactive card (its just going upstairs, its less interactive than any vanilla creature), but boros charm is sometimes interactive, so maybe thats enough to put it into the second category. Bolt is somewhere between threat/nonthreat depending on the deck as well.
This also somewhat solves the tron problem.
In theory, ugin is an interactive card, but in practice, its a coinflip of a few bolts/EEs or just straight up wins the game because the other player can't possibly beat it, so it should be worth less points than O stone, but more points than urza's tower.
I shouldn't need to explain the difference, and I'm a little puzzled why I have to. Moreover, it's particularly unfounded and bizarre that you are challenging the entire proposed system without seeing any examples of it applied, and when I literally said it was still a work in progress.
1. Such a system would transparently define all terms and ratings and would be fully auditable by readers. Hoogland just grouped a bunch of stuff and didn't bother to explain the terms or groupings.
2. Such a system would be applied across multiple eras of Modern to test for false positives/negatives. Maybe even across formats. Hoogland just loooked at a handful of T8s and made a sweeping format generalisation without comparing his claim to other time frames if it's even a valid analysis method.
3. I am anticipating objections and flaws with the method. Hoogland just took it as gospel in revealing some big Modern reality without acknowledging any other competing interpretations.
These are just a few differences that should be obvious in just the post I wrote about this. Most rating systems start out as subjective, but if they actually measure what they claim, they become much more objective measures. ELO is an excellent example of this. If the proposed system does as promised, great, we can use it. If it makes too many false positives and negatives, we'll try to refine it or move on.
@Others
Thanks for the constructive feedback. I think those are interesting starting considerations. Matchup context is particularly interesting, but might be too complicated. More work is clearly needed!
As a math teacher who uses Excel spreadsheets for fun, the idea of doing this seems fascinating. But as a useful tool for anything other than an interesting thought experiment? Not as much. Have fun with it. I know I would.
Edit: Speaking of interesting data sets; https://www.reddit.com/r/ModernMagic/comments/aazife/how_long_games_last_in_modern_97674_mtgo_games/?utm_source=reddit-android
UR ....... WUBR ........... WB ............. RGW ........ UBR ....... WUB .... BGU
Spells / Blink & Combo / Token Grind / Dino Tribal / Draw Cards / Zombies / Reanimate
So all rating/ranking scales are just subjective and useless thought experiments unless they rate/rank purely objectivr, ordinal numbers? We can't use any quantitative systems to measure or capture seemingly qualitative phenomena? This is the sort of reductionist logic which in a few sentences dismisses the foundation of most sports, social science, political, and economic scales we use. Even predictive web search ranking algorithms seem to just be useless thought experiments given your attitude. As such, I don't find it particularly credible. If a system is valid, reliable, transparent, replicatable, and meets other standards, it ceases to be "subjective" in the dismissive way you suggest and becomes a valuable analysis tool. Note that Hoogland's absurd pie chart with its extremely opaque categories does not attempt to meet this bar.
If I had posted the system with examples and it did not meet those benchmarks, I would understand your challenges. Given that I have not done that and you have already rejected it as not being a useful tool, I find it pretty easy to determine your objections are at best premature, baseless, and unhelpful.
EDIT: that MTGO analysis is a perfect example of a rigorous, data-driven approach to answering the question of format speed. Note that Hoogland's terrible analysis of 66% of decks trying to win/set up an insurmountable game state by T3 is largely rebutted by this graph, which shows that only 8% of games end on or before T3, and only 22% end on T4. Who cares what decks are trying to accomplish by some subjective standard when it cant even remotely predict the turn that games end on?
Regardless, I don't know who could look at either of these and think: "You know, I think I want to play a deck whose win condition is a planeswalker ultimate on turn 10, or a 4/4 that I can't activate until turn 6, but realistically don't activate it until 10+."
UR ....... WUBR ........... WB ............. RGW ........ UBR ....... WUB .... BGU
Spells / Blink & Combo / Token Grind / Dino Tribal / Draw Cards / Zombies / Reanimate
But who cares when decks are capable of winning? How is that meaningful in any way? It's even more meaningless when we have no clue how Hoogland defined those decks or even what decks are in that percent. It is significantly more valuable to know the actual turn distribution that games end, as this far more useful analysis shows.
The moment I'm home, I will see how this 2018 distribution compares to the distribution in 2015 Modern, which I know many view as the diverse golden age of multiple pillars with a low incidence of linear decks. That will allow us to see how Modern's speed has or has not changed since 2015 and the year of Twin, BGx, Bloom, and other 2015 mainstays.
The musing just got more annoying to say it this way
Greetings,
Kathal
Modern/Legacy
either funpolice (Delver, Deathcloud, UW Control) or the fun decks (especially those ft. Griselbrand)
If you're looking to make a deck choice decision, are the two really any different in their conclusions? Since I edited my last post to include an extra line, I'll paste it here again, since it directly addresses the practical uses of data like this: I don't know who could look at either of these and think: "You know, I think I want to play a deck whose win condition is a planeswalker ultimate on turn 10, or a 4/4 that I can't activate until turn 6, but realistically don't activate it until 10+."
UR ....... WUBR ........... WB ............. RGW ........ UBR ....... WUB .... BGU
Spells / Blink & Combo / Token Grind / Dino Tribal / Draw Cards / Zombies / Reanimate
They might say "Those other decks seem ill-equipped to win the long game, so if I can tune my answers to keep me alive until T5 I have a really good chance of leveraging stronger late-game win conditions." Both of those hypothetical and rhetorical positions are equally plausible.
Incidentally, as Kathal predicted, there is a very small difference in the turn distribution between 2015 and 2018 Modern. See the numbers below. An earlier version of this list combined the T0/T1/T2 numbers but I realized separating them is clearer. I'm not really sure what a T0 win is if neither player has a turn (seems like someone just forfeited before cards were played), so I might need to just remove that line entirely because it might not make sense.
2015 vs. 2018 cumulative win percentages by Turn N
T0: 0% (2015) vs. 0.5% (2018)
T1: 0% (2015) vs. 0.9% (2018)
T2: 0.9% (2015) vs. 2.4% (2018)
T3: 6.2% (2015) vs. 8.3% (2018)
T4: 21.4% (2015) vs. 22.2% (2018)
T5: 39.4% (2015) vs. 39.4% (2018)
T6: 55.7% (2015) vs. 55.2% (2018)
T7: 69.4% (2015) vs. 67.6% (2018)
T8: 79% (2015) vs. 77% (2018)
T9: 85.5% (2015) vs. 83.9% (2018)
T10: 100% (2015) vs. 100% (2018)
The biggest points of difference are in the T0-T3 wins and the T9/T10+ wins. In 2018, 2.2% more games end on T0-T3 than in 2015, which indicates a slight increase in early game speed. That said, 2018 also sees 2% more games go beyond turn 9+, which means the long games go longer than they used to. This ultimately results in no difference between the average turn on which games end: using weighted averages, that would be turn 6.43 in 2015 and turn 6.42 in 2018.
Greetings,
Kathal
Modern/Legacy
either funpolice (Delver, Deathcloud, UW Control) or the fun decks (especially those ft. Griselbrand)
Yep, thanks! I just posted my own analysis follow-up to that thread:
https://www.reddit.com/r/ModernMagic/comments/aazife/how_long_games_last_in_modern_97674_mtgo_games/ecwsyni
I am really curious to see how this distribution compares to Standard, which I know he is also working on. I suspect Standard will be slower but only by 1-2 turns, at least if my MTG Arena experience is any indication (which it might not be because I predominantly play BO1 which has different dynamics).
I wish I had the patience to go and find all the posts and comments about how banning Twin would slow down the format and allow you to tap out on turn 3!
UR ....... WUBR ........... WB ............. RGW ........ UBR ....... WUB .... BGU
Spells / Blink & Combo / Token Grind / Dino Tribal / Draw Cards / Zombies / Reanimate
Has anyone ever argued that banning Twin slowed down the format? I don't recall those arguments, and if I ever personally made that argument, I chalk it up to temporary insanity. Banning Twin obviously ran the risk of speeding up the format. I'm sure I've even acknowledged that in the past. We just didn't know how much it would speed up the format, if at all. But now we have numbers to assess how much faster the format has become following/as a result of that ban.
Tapping out on T3 does not impact T0, T1, or T2 wins (you can still tap out against Twin on T1/T2 with no fear of Twin winning the next turn). So that means we just need to look at the relative difference in T3 wins to see the win-turn difference in Twin and no-Twin Modern. In 2015, 5.3% of all games ended on T3. In 2018, it was 6%. Even if we can chalk all 100% of that .7% difference to Twin, which I doubt we can, that's still a pretty tiny margin. It's also not a very persuasive case to unban Twin. There are definitely good arguments to support an unban, but claiming that unbanning Twin will slow down Modern to the tune of .7% of games is not exactly compelling.
Perhaps not directly, but at least one of the things dozens of people were saying how much more playable 4-drops would be as a result of Twin being gone, implying that larger cost spells on curve were more viable due to not dying quickly.
I remember this vividly because people have been saying all sorts of stupid and untrue things related to Twin for at least 3 years. And I've been there the whole time listening to it and dealing with it.
UR ....... WUBR ........... WB ............. RGW ........ UBR ....... WUB .... BGU
Spells / Blink & Combo / Token Grind / Dino Tribal / Draw Cards / Zombies / Reanimate
This plays into my earlier comments about counterspells, and how turn 2 (on the draw) ones play a different role to turn 3+ (on the draw) ones, and how Absorb will likely be competing for slots with Cryptic Command, rather than the 2 cmc stuff. I really hope WotC starts pushing some better 1-2 cmc main-board worthy answers into the format and push those turn 3 or earlier games down below 1% overall, and only around 5% for specific faster decks (which would win turn 4 on their faster games most of the time outside of nut draws and the opponent drawing terribly despite mulliganing as much as is worth the risk).
Just a little input on it. Not really debating anyone. When Twin got banned it's hard to say if the format got faster or more slow. Right after the ban people here in our area became more complacent tapping out on turn 4.. slower aggro decks without counterspells like Elves became more common. Well, decks like Affinity, Living End, and Burn still tried to win as fast as they can.. but you can see that they're no longer anxious tapping out turn 4 against anything with red and blue in the manabase. When twin was alive, those decks often have something like a galvanic blast, beast within, or combust ready to kill a Deceiver Exarch.
Nexus MTG News // Nexus - Magic Art Gallery // MTG Dual Land Color Ratios Analyzer // MTG Card Drawing Odds Calculator
Want to play a UW control deck in modern, but don't have jace or snaps?
Please come visit us at the Emeria Titan control thread
-Eldrazi Winter
-Summer of Dredge and Infect
-Shadow (slowed down)
-Humans
Right now we're maybe in a Phoenix meta? Phoenix may actually slow down the meta, as it's weak to combo decks and looks soft to decks like GBx/GDS.