A self correcting variant Netrunner format

There’s been some talk about playing Netrunner with some different sort of card legality that fills the void left by glacial rotation, or the permanent lack of core set rotation. BB45 is a cool movement. After my experience with another competitive game with a small community though, I have a sort of fear and anxiety about customized rebalancing rules coming from players within the community itself. It’s hard - impossible actually, even with efforts in the best of faith - to be unbiased. You view the game from the lens of your own experiences, and no one plays equal proportions of Shaper, Anarch and Crim (if you do, you might want to consider stopping that, your performances tend to be better if you practice one type of deck extra.) I’ve seen players with sway over a ruleset, either for personal gain, or because it’s impossible to to divorce an assessment of imbalances from an assessment of experiences in which you were outplayed by an even better player, restrict or ban strategies that didn’t really warrant it.

In a CCG with the unique property that most players own every card (and in a variant format, proxying would be allowed anyway), I like the idea of a dynamic, self-modifying (pun intended) banlist that simply views tournament results from the previous season and restricts or bans cards accordingly. A cold, unfeeling algorithm could make that decision, simply removing the cards that appear in a disproportionate enough of victorious decks (or adding some restriction to those cards.) This removes possibly unfair kneejerk bans, like banning Accelerated Diagnostics after one season when the underlying cause was actually that the CI players had been practicing their Netrunner more than the players learning how to fight the deck, and perhaps CI was not strongly spread as an autowin option amongst the top 50% of the tournament, as a sample hypothetical example.

A downside is that the algorithm hits the wrong card in a strategy, say, Adjusted Chronotype gets restricted or banned for high usage, but some poor Sunny using it with Globalsec Security Clearance has to learn a new deck now. I think that’d be pretty rare and an acceptable loss. Most of the time the essential aspect of what makes the strategy good will still be there.

One issue I imagine when I think about the concept would be an issue of “flicker”. If cards don’t stay banned forever once the algorithm bans them (it seems sad to say goodbye to a card forever, although if the algorithm bans cards more slowly than new cards are released, you would in fact still have an ever-growing game), then very strong cards would pop in and out of legality. A card would be banned, after X amount of time it unbans again if you don’t go with the permaban option, everyone might know that it’s gonna be a problem, it’s a problem, that particular season is pretty stale, it immediately goes out again. Restricting the number of instances of the card could reduce flicker in some cases, but in others might not help enough. A card like Eater that needs to be 3x for it’s strategy could easily flicker hard on just whether the 3rd copy is allowed.
Does anyone have any ideas about the flicker issue? Amongst those that don’t hate the idea entirely, which I’m sure will be most of you!

Thanks for reading yet another silly popsofctown thread. I appreciate your feedback, even if it’s negative :smiley:

3 Likes

I dream of a day where the cardpool can be self correcting in such a manner - ideally by the costs of the card changing… (e.g. sure gamble starts costing 6 to play to return 10?, desperado increases to 4, then 5). In theory, the MWL could be used in this manner… with possibly an added dimension of how many influence “pips” they are taxed - which could address the “flickering” impact (you could use partial pips, maybe desperado could go down to 1/2 or 1/3 pips)

1 Like

You might be interested in this tournament that a group ran about a year and a half ago. After each round of the tournament they banned the 18 most popular cards (2 of each non-ID type, e.g. Assets, Agendas, Programs, etc). Not really as “proper” of a format as you are imagining though – the algorithm is pretty dumb and irreversible, and it was intended to devolve into silly situations as the weeks went on.

2 Likes

I don’t think a CL2 category would help a lot.

For exemple for programs, you could put a tutor and have the same result minus a clic and/or a few creds in 1/3 of your games ?

Individual cards does not pause a problem to me. The main problem are “unfair” archetypes that abuse synergies.

For me, MoH isn’t an issue if you play it oo Genomics. Faust isn’t an issue if you play it oo Wyldcake decks.

We could also say the game need more rules like “one cannot shuffle more than once a turn” or putting Faust on MWL, but I’m not sure this is the main problem.

The main problem seems to be : testing is not made in good conditions. The game need good deckbuilding testers that try to deviate a card from its purposes. They are not deviating cards enough.
I don’t know if testers have access to 20 cards here and there, or whole cycles too.

But dangers should be easy to spot : tutor, recursion, ai, 3/2s, blue/yellow FA, green tag etc all are nuclear stuff.

Hypothesis:
Make the banlist be graduated. First pass, those deemed ‘most oppressive’ by the algorithm are restricted, one less copy in your deck is allowed to be played. Every pass, the ‘most oppressive’ cards get restricted one more card. It’s possible that if a card is so oppressive that it makes this list even at 1-of, then it could be banned entirely. Every X passes (I think either 3 or 5 for X would be good to start) if a card hasn’t increased its restrictiveness, reduce its restrictiveness by one.

Mostly solves the ‘flickering’ issue that you noted, though as long as you have an algorithmic reversible ‘banlist’, flickering will happen at some level.

The hard part is determining ‘most oppressive’. As a starting point, I’d say any card that appears in every deck that makes the cut of a Regional or Store Championship probably qualifies. A card that appears <MAX_ALLOWED> times in every deck that wins a Regional/Store Championship probably qualifies. (Difference is that it could be a singleton in the first case, but it’s a full playset in the latter.)

The main point and thrust of a banlist like this is to try and solve play diversity. It doesn’t help with Game Design Space.

1 Like

Right, it definitely won’t do the sort of thing like restricting Yog so that giving a code gate 3 strength instead of 1 strength has some sort of niche or meaning once again.

But I don’t know, maybe it’d be enough, maybe it’d be fun, and maybe flicker wouldn’t be such a big deal. I would worry that certain cards would have even worse NPE if they are at 1x, though. Making “correct” runs to hit a 1x NAPD while on 3 credits, leaving HQ open and losing to singleton siphons, having Sure Gamble in an opener and then beating a runner that just didn’t have any econ in their opener.

If you had detailed records of Netrunner games (i.e. because you had a strongly rules-aware online play platform), you could start estimate some metrics for any single card C:

  1. Probability of victory, given drawing C
  2. Probability of victory, given drawing C on turn t (needs more data, but more valuable than 1)
  3. Probability of victory, given playing C on turn t.
  4. Probability of victory, given playing (a copy of) C k times.
  5. Expected number of turns between drawing C and playing C (this implies that C is situational or part of a combo)
    5b. Expected number of turns between drawing C and playing C, given the presence of some other C’ on the board (this heuristic is helpful for identifying other combo pieces)
  6. Expected number of turns between playing C and the conclusion of the game

With these heuristics, you could start to get a handle on defining “oppressive”, e.g.:

  • If playing C early almost assures victory, but the expected number of turns until the conclusion of the game is large (this implies that C is part of an overpowered engine).
  • If probability of victory increases monotonically in the number of times k a card C is played (spamming the same effect over and over is uniformly better than doing anything else).

You could then, perhaps, calculate some sort of “oppression score” for a deck, and restrict on a deck basis, rather than a card basis. NRDB and Meteor could consume this score as an API, and present the user with which cards in their deck are most oppressive or something.

1 Like

There was a huge database of crazy awesome data like that for Dominion when it was played online on isotropic, it was called Councilroom. It could determine which cards good players were more likely to use, whether having a card made you more likely to win, whether one play having the card and another player being a naysayer on it indicated that one player was probably worse at the game. It was awesome.

1 Like

Just think of the optimal opening hand for Wyldcakes and ban that