25 Oct, 11 tweets, 3 min read
Random thought from some Discord discussion: one problem with how some people use 17lands data, that has existed before 17lands though things like hypergeometric calculations, is that people like quantifying things, but often don't actually connect numbers to actual meaning. 1/x
Example: let's say I'm trying to decide whether to keep or mull a 2 land hand, and calculate that I have an 85% chance of drawing my third land by turn 3. One might say "85% is pretty good, I'd keep". But an annoyance I have is that I don't really know what this means! 2/x
How do you know that 85% is "pretty good"? Sure, it's a relatively high number, but I'd bet most people who might make such a statement don't really think through what the number means, and just like that it's high. At what number would you mull the hand? 80%? 75%? 70%? 3/x
Is a 15% chance of missing my third land drop an acceptable risk, better than the risk of mulling? Did you take into account the situations where you hit your land drop only one turn late? What about if you hit your third land drop, but miss your fourth? 4/x
My point here is that Magic is extremely complex and intractable, so it is often very hard to calculate anything of strategic value. The best you can do is usually through comparison - for example, Karsten's charts of how manabases relate to castability of spells. 5/x
But even that can often fall short. Instead, strategy is developed through intuition and practice - people try things, and see what works. People don't keep hands because the chance of hitting their 3rd land is 85%, they keep because they've had success keeping those hands. 6/x
This is not to say that simple models and data are useless - in fact, they can still be very useful! Calculating metrics through which to categorize different situations can be very helpful in helping develop strategy - maybe 85% *does* mean something to some people. 7/x
"Card advantage" is one of the simplest and best examples of this. I can calculate how many cards my Arcane Infusion draws on average given 10 instants and sorceries in my deck, and I can use past experience to know that drawing 1.4 spells on average is a pretty bad rate. 8/x
But it is very important to recognize when you're actually properly using numbers to your advantage, and when you're just recognizing big ones. And it's important to be able to communicate this. 9/x
Conclusion: Magic is an extremely complex game, so don't expect to be able to glean useful direct strategy from straightforward models and calculations. Instead, use these simple models and data to compare and categorize situations, in order to help develop your strategy. 10/10

• • •

Missing some Tweet in this thread? You can try to force a refresh

This Thread may be Removed Anytime!

Twitter may remove this content at anytime! Save it as PDF for later use!

# More from @JasonILTG

25 Oct
I know a lot of people like making tierlists for Storybook Brawl heroes, but I really don't like tierlists, so instead here's a thread with some thoughts I have on the strengths and weaknesses of every hero in SBB, sorted by my avg placement with them on this patch: 1/x
First up: I have not played Evella, Gepetto, Jack's Giant, Krampus, Mrs. Claus, Pan's Shadow, Pup the Magic Dragon, or The Fates on this patch, but they all seem like fairly medium midgame stats heroes to me. Some are probably better than others, idk. 2/x
I also haven't played Grandmother on this patch, but from previous patches, I would treat it similarly to the above midgame stats heroes - one of the main benefits of midgame stats is saving HP, and 10 extra HP is a lot; plus, +3/+3 is big enough to be relevant lategame. 3/x
24 Oct
Made it to top 32 in the #SBBTourney, which should be starting soon. Streaming my view at twitch.tv/jasoniltg with a 15 minute delay!

You can also find the official coverage at twitch.tv/storybookbrawl and the leaderboard at sbb-tournaments.netlify.app/2021-october-l…
Looks like I'm in the featured pod too!
Unfortunately starting with a 6th with weak hero choices - never really leveraged Charon well, starting on a Happy Little Tree but not finding further scaling to run it, and not getting a critical Poly slay on 4.0. Will have to place extremely well in the next two to get top 8.
24 Oct
Streaming day 2 of the #SBBTourney with 15 minute delay, starting soon! (Should have mic this time; hopefully the delay works)

twitch.tv/jasoniltg
Oh also - you can watch the coverage here: twitch.tv/storybookbrawl

And you can follow the standings here:
sbb-tournaments.netlify.app/2021-october-l…
Starting off with a first is very nice! Very strong early start, leveraged with Piggy Bank plus kissing a Brave Princess three times to get Ashwood+ while on lvl. 4. Cat's Call let me put the nail in the coffin against Medusa Mirror Mirror Hoard Dragon.
24 Oct
Streaming the #SBBTourney now! (5 minute stream delay, mostly not reading chat)

twitch.tv/jasoniltg
First round is a Pants first! Got hatball pretty early (I think it took 3 or 4 lvl. 3 triples?). Never found Fork/Wand, but it was enough.
Second round is 5th with Mad Catter - strong midgame and got a bunch of random 6s from Ogre Princess+, but didn't get anything coherent and died with Herc+Treasure Map about to go off
17 Oct
Very interesting video from jorbs (excellent strategy game streamer who mostly streams Slay the Spire) on how he thinks about strategy games. I would highly recommend watching the video yourself, but I figured I'd do a thread connecting some of his ideas to MtG draft. 1/x
Again, want to stress that I'd highly recommend watching the video first! jorbs uses a lot of Slay the Spire examples, but the concepts are applicable to all sorts of strategy games (and life in general). This thread is just me applying those ideas specifically to MtG draft. 2/x
First, a small overview: the main concept jorbs talks about in his video is what he dubs "loose Bayesian knowledge" - he approaches strategy games by having prior expectations and hypotheses that he updates through his experiences playing and experimenting in those games. 3/x
26 Aug
The latest Lords of Limited episode was absolute 🔥; fantastic discussion by the lords and twoduckcubed about how to best use 17lands data. However, one thing that I couldn't help but notice was missing was my most-used 17lands stat: Average Last Seen At. So here's a thread! 1/x
A definition: Average Last Seen At, or ALSA, is: "The average pick number where this card was last seen in packs. When a card comes back around on the wheel, only the second time around counts toward the average." So basically, the higher the number, the later the card goes. 2/x
One of the benefits of ALSA is that it is inherently less complex than most stats. GIH WR will be aggregated across a large swath of decks, gameplay situations, and play skills, while the only thing ALSA's aggregation hides is information you might get from signals. 3/x