QuantANR: Analysis of a Changing Meta

Caution, Wall of Text ahead.

When it does get to that stage, we can talk about how the factions differ (with criminals having runs costed more in terms of achieved draw quality up to that point, for instance). For now, I’d just assume assembing a rig will cost the runner about 10 credits, spread evenly over the first 6-8 turns or so. 10 credits feels about right for both Shaper and Anarch:

  • Mimic, Sucker, Corroder, two Parasites for Code Gates
  • Lady, Cyber-Cypher, 4 creds to get through Sentries

Criminals are a bit trickier (damn you, Fairy and Inside Job), but when you factor in the install clicks that could have been resource clicks, I’m guessing you come in at around 10 cred as well.

The most tricky ones to compute will be the circumstantial ones (because we can’t properly evaluate when to add an additional cost to them), and the one-use ones (because there we need to count the clicks spent installing more often). Also, stuff including heavy Parasite recursion has good potential to screw the calculations up.

Co-simulation, while being the desired final state of this experiment of course, might be overkill for what we need right here and now. I’m thinking more along the lines of “add an estimated sum of credits needed to interact with the corp at various moments”, much like we did with costing the runs. My gut says that while this money will be spent in various ways against various corps (trashes vs. NBN as opposed to more expensive barriers vs. Weyland, for instance), the overall sum is going to remain more or less the same.

Thinking about various corp decks, here’s my stab at abstracting the interactions into a general cost estimate:

  • Not wanting to lose will probably cost you around 5-6 credits over the course of the game:
    • Carapace for 3, clearing an inopportune Snare tag once
    • Deus X twice (to eat, say, a Snare and a Ronin/Philotic)
    • Non-kill decks might well cost you more than kill decks on their closing methods (repeated runs on HQ trying to fish out the agenda so they can’t Biotic it), beating Ash traces, running the server multiple times due to Caprice, etc etc), so in general I’d tend to feel safe adding this 5-cred buffer to @Nordicstrike’s initial calculation for any opponent
  • Given the prevalence of NAPD Contract (and to a lesser extent TFP/Fetal/must-trash assets), I think we should assume you want to have at least 4 credits on hand when you’re done getting into the server, and that you’re not running until you can do that. The run won’t always cost you those credits, but you want to have them on hand (and spend them, say, 25% of the time, randomly)

I think just adding these two things would give a significantly improved result.

So, what you’re saying is… they’re not any more? Damn, I gotta step up my game! :stuck_out_tongue: (also, thanks!)

edit (damn you, NordicNinja): I don’t buy the 7 credits. SMC costs 2 to use, Chip costs whatever you’re pulling back. Other than that, we seem to be thinking in the same general direction, though!

1 Like

PPVP has highest priority. DJ < LF < SG. This is obviously not always going to be appropriate, but it 1) optimizes for efficiency (PPVP earliest), and 2) would result in the most flexible use of clicks (SG > the rest means more clicks left to do other things). Good question!

1 Like

Hey, I just want to say thanks for your time and effort @YCombinator. This is really good stuff that can help new and experienced players alike. I’ve been teaching a friend how to play and showing him your data is more convincing than saying ‘just because’. Keep up the good work.

2 Likes

Sorry for the hiatus. It has been a busy week at work.

This is hilarious. I’m optimizing on the code for simplicity, and I doubt the core will grow over 1K LoC, so I’m not too worried about this. However, taken to an extreme, complexity and intractability would obviously be problems. I oversee a 250K LoC code-base, so I’m pretty allergic to complexity.

Fair enough. I think that doing the simple thing will pay off. If it doesn’t we can go to the more complex option.

I actually meant here that I wonder which decks are well known but have consistency issues. I’d like to quantitatively show how they compare to the “tuned” decks, and zero in on why they are inconsistent. This is essentially a variant on the scientific method: we as a community have a hypothesis about specific deck constitutions. Can I confirm those hypothesis? If not, then there’s a problem either in the assumptions, or in the simulation.

Re: some program costs and NAPD/taxing runs. Cool. This, again, might be a simple short-term solution, and we’ll see if anything else is necessary.

You and about three others are still at the forefront for me ;-). Don’t worry, you’re still doing a good job.

Thanks for spending the time to read it! I’m glad you got something from the article.

1 Like

In that case, I think the best to start comparing is a standard-ish Wyldside/Aesop’s Noise vs. an AndySucker, as those are pretty much the polar opposites in terms of consistency.

In general, when looking for a well-known but inconsistent deck, just look for something with a 3-of that’s utterly critical for the deck to get off the ground, combined with very little draw and/or tutoring for it. Steven’s AoA would be another good example, as without getting an early Siphon to connect you’re usually just hosed.

(overall, any pre-Inject Anarch will probably serve you well as an example, come to think of it)

I think it will be a little while before I can approach Noise and Crim. Card inter-dependencies are annoying. I agree they will probably show an interesting aspect of this. Thanks for the ideas!