A New Model for Card Analysis

What’s the biggest LCG or CCG by player count? Magic: The Gathering. I didn’t look that up. Didn’t need to. Because it’s not just the biggest, it’s the biggest by a stupidly large margin.  And since that’s where the money is, that also means that the analytical understanding of Magic is much richer than any equivalent game.

Problem is, Netrunner is a completely different beast than Magic, and one of the biggest differences between them is where the complexity lies. Magic the Gathering has highly complex deck building. Netrunner has highly complex options in game. This is a very fundamental difference, and it means that most of what we’ve learned about card analysis from Magic (which has deckbuilding hundreds of times deeper than Netrunner can ever hope to have) simply doesn’t apply in Netrunner (where the deckbuilding is simpler, but the cards exist in a much richer context than in Magic).

My thesis is, essentially, that this is hurting our analytic understanding of Netrunner.  Not directly – no one is taking concepts from Magic: The Gathering and applying them wholesale to Netrunner. But we are taking it as a de facto assumption that we can analyze “good” and “bad” cards by looking at how “efficient” or “strong” they are – because that’s true for Magic. In a game of Magic, cards exist mostly in a deck context. But in Netrunner, cards exist mostly in a game context. It is my belief that meaningful analysis of Netrunner must make allowances for this difference.

Context and Interactivity

Let’s look at another question that’s been raging at the border of Magic and Netunner – which game is more interactive?

Well, it depends.

In terms of things you do on your opponents turn, Magic wins easily. And that kind of seems like interactivity, doesn’t it? You’ll have a lot more direct foils to your plans – more counterspells, more Wrath’s – in Magic than in Netrunner. There are a lot more possible ways for you to interact with your opponent.

But….will that change your play algorithm? If your plan is to turn a bunch of dudes sideways, and you’re playing a turn dudes sideways deck, and one of your dudes is counterspelled – you don’t suddenly try for a control win. Magic doesn’t work that way. You’re still trying to turn enough dudes sideways that the loss of your dude doesn’t make you lose. In other words, Magic has a lot to do to someone else, but what YOU do from game to game doesn’t depend much on what the other person is doing. (I’m using the relative much” here. Obviously, skilled Magic players do exactly this, and I’m not saying that Magic doesn’t have an element of playing around your opponent – merely that it pales in comparison to executing your deck’s “game plan”.)

Making Magic decks requires a lot more thought and effort than Netrunner. Unless Netrunner adopts instants and the stack, that’s never going to change. The “plan” of a Magic deck can often be extremely complex. But in one field – the game-to-game variance in the decisions you make using a deck – Netrunner wins hands down, no contest. THAT is the way where Netrunner is more interactive than Magic.

There are two major reasons for this. First off, the click system exponentially expands the decision space. In a Magic deck, you can play the cards you have the mana for and use the things you have. In Netrunner, you can draw cards, or get creds, or make runs, or install…In terms of simple mathematics, a Magic game CANNOT offer more choices than a Netrunner game. The other reason is a bit counter-intuitive. It’s that Netrunner games are forced to be similar. I know the difference between, say, an aggro Gabe and a big dig Whizzard may seem massive, but it’s a drop in the bucket compared to Magic. And because of this, your opponents choice will always be a major factor in your decision. In Magic, the decision to make an Aggro deck is imprinted so fundamentally in the deck architecture that even against a deck that is massively vulnerable to Control, you’ll just be attacking anyway. But in Netrunner, any runner can ALWAYS make a run,  and a Corp can ALWAYS score an agenda. What this means is that it’s always possible to exploit your opponent – even in a dedicated Scorched Earth Weyland deck, if your opponent never gets a way past your Ice Wall, you can score agendas and never SE at all. Even in a dedicated Noise Mill deck, you can just run R&D and never hit archives if R&D never gets Ice.

To summarize: imported ideas of card analysis will probably focus too much on the deck as the unit of analysis and not enough on the gamestate, because in Magic, your deck relies mostly on sideboarding to change how it operates against a particular opponent.

It is with this consideration in mind that I propose a new model to analyze cards in Netrunner. It’s far from perfect, because my understanding of the game is far from complete. It’s also much LESS quantitative than most existing models. This may make it seem less usable for card analysis – and that’s true. It gives less raw data, creates no graphs. But I believe this is an inherent property of Netrunner – hence the above – and I would rather be generally correct than precisely wrong.

Introducing: The Click Market

“Android: Netrunner is a game about maximizing the utility earned from the clicks you spend on the click market.”

There’s my idea in one sentence. Well, that doesn’t explain much. It’s a good thing there are still some words under this one.

The first thing this does is establish clicks as the currency of Netrunner. Many people looking at the game make the excusable mistake of thinking creds are currency. The problem is that this model can get you in to all sorts of trouble looking at “economy” cards. The fundamental premise behind looking at economy cards tends to be: “Given a set number of clicks, which options make  you the most money?” But the number of clicks isn’t set at all! Credits are  a secondary resource, a necessary evil for other options that win you the game. Instead, in Netrunner, you’re spending clicks and trying to win the game. Now, you’ll often need to spend clicks on an option that gives credits, because those credits make it possible to spend clicks on something ELSE that will help you down the line. I’m not saying credit cards are bad, or anything silly like that. What I’m saying is that when most people look at money cards (they tend to be the focus of quantitative analysis more than anything else, because it’s here that people get most confident in their models) they take it as a de facto assumption that your goal is to make “the most money” – but your goal is to win the game! You’re not allocating some set number of clicks to credit generation. You’re spending your clicks on whatever is most valuable at the time, which will often be credits, because without them many other choices have no value.

As a couple of preliminary notes, this means that efficiency analysis will undervalue burst economy (it’s not worth much in terms of clicking for cred instead – but it’s worth a lot for short term spikes in utility) and, similarly, limited-use economy options that “finish out” quickly – because even if their click to cred ratio is lower than other options, spending fewer clicks on cred generation at ALL is a benefit. Efficiency analysis will also overvalue things like Kati Jones that have gating on when they can be used – because even if you’ll spend one click in four on cred generation on average, FORCING it is a very real cost that doesn’t appear anywhere in credits divided by clicks equals value. (Kati is still a good card.)

You may be saying: Cost in what? Value in what? In utility. It’s a vague term. That’s deliberate. It’s so no well-meaning blogger tries to quantify it. You can speak of utility in relative terms – an option has more utility, or less utility. But there’s no basal unit that you can divide or multiply or whatever. This is because utility in Netrunner is dependent on the gamestate. Utility is how much something helps you win the game. Creds have no inherent value – utility has value. Creds are just things that can let your clicks chose options that give more utility.

Let’s get to that last bit, the “Click Market”. What’s that? A concession to the fact that in Netrunner, you can do whatever you want. Trying to treat the game as an excercise in maximizing creds is doomed to fail – the optimal solution is clearly to get Magnum Opus and ping it four times a turn until you lose. That’s why we needed “utility” (barely) defined as the thing we’re trying to get the most of. And clicks are the currency we’re spending to get it. The click market is the interface between the two. It’s a place where click-vendors hawk their wares. You can get a credit. You can get a run on R&D. You can get a card installed from your hand. Some cards, like Opus, give you a whole new option (with almost strictly more utility than another one.) You spend your clicks on the click market and attempt to maximize the utility you get for them. That’s Netrunner.

Every option on the click market has a corresponding utility value. These utility functions have many inputs, which we’ll simply call variables. Credits feature prominently, but MU, Link, and space in a remote server also qualify. There’s also the number of other clicks in your pocket – the Tinkering option is worthless if you’re spending your last click on it. There’s the number of cards in each deck. There’s the credits of your opponent. The entire game of Netrunner can more or less be described as a game of how good you are at judging, based on the variables, the output of the utility function of each option. And every click you spend, the market fluctuates. This is the point that most current analysis misses. People looking at economy cards assume that the value of one cred is constant – when in reality, the utility of one more cred is predicated on many other things.

(A note: we’re maximizing our utility across the whole game, NOT picking the option with the most utility every time. For example, installing a Gordian Blade has zero utility. What it does is increase the value of “Make a run” when there’s a Code Gate installed. When we install Gordian Blade, we’re hoping the waste of a click and the spending of secondary resources will be compensated by the fact that running on, say, Enigma has utility it previously did not.)

Take the option “Click: Spend one credit, place an advancement token” and look at it’s utility function. We can’t quantify this, but we can tell which inputs are major, minor, and irrelevant. And one of the most significant in this function is the “remaining difficulty” of each agenda on the field- that is, the difficulty of the agenda minus the number of advancement tokens currently on it. SanSan City Grid  is a good card because it lowers the RD of an agenda in its server by one, and the utility function spikes up massively in value when an agenda has an RD of a value equal to or less than the number of clicks in your pocket. Why put SanSan in your deck, when credit-wise, it’ll almost never save you those six credits? Because it will greatly increase the utility of an option you intend to utilize during the game, either by directly giving you that option (Magnum Opus) or (as in the case of SanSan) manipulating a variable or variables important to the utility function of that option. That is what you’re looking for when you’re picking cards for your deck.

I chose this example specifically because it’s something everyone already knows. The good thing about SanSan is scoring 3 for 2’s from hand – that’s not anything new. No one attempts to model SanSan based on it’s “efficiency”, because it’s obvious that there’s far more to the card than that. The problem is that some people who know that SanSan is unquantifiable still assume other cards are quantifiable. We can intuitively understand that we’re taking SanSan for what it does to this utility function, but it’s harder to see that Easy Mark giving you three creds for one click might mean more than just the value of 3 creds, minus 1 cred for the option you could have chosen instead, will give you – because the point of getting creds isn’t just to get creds, it’s to make other utility functions give higher outputs, and getting 3 creds in one shot does this very effectively and quickly. The point isn’t to compare a credit option with clicking for cred over an arbitrary timescale – it’s to look at EACH credit option, and think “How will this influence other utility functions?”, and choose what to include based on that.

This also takes care of that venerable old beast “You forgot the click to draw.” It’s one of those non-thoughts that’s seductively close to meaning something. I mean…you could have clicked for cred twice instead of drawing and then playing that Easy Mark, right? The problem is that this is comparing the COST you spend on each option, implying an equality. But that’s literally the opposite of the truth! The entire point of the game is to choose the options with the most value given a set number of clicks – and there’s certainly no guarantee that gaining two creds was worth two-thirds the utility of the Easy Mark. Instead, the utility of your draw function depends on what’s left in your deck, so the “cost” of the click to draw depends on what your draw looks like. Similarly, the equality “one click is worth one cred is worth one card” is wrong, wrong, wrong. The correct answer is “One click can get you various things, pick the best one.”

Drawing is not some sort of “pre-cost” you pay on an Easy Mark. Instead, the utility of your draw is one thing, and the utility of the Easy Mark is another, and that’s that. What putting an Easy Mark in your deck means is that sometimes your draws will give you an Easy Mark, which will affect your draw function. When we analyze Easy Mark, we don’t say “Given the click to draw, playing Easy Mark is one cred better than clicking for cred.” We say, “Easy Mark influences our draw function negatively when we’re looking for other cards [this is true for every card, and is the true cost of “click to draw”], and influences it positively when we’re looking for creds.” When will we be looking for creds? When we’re poor. Then a quick burst of creds is great. So, Easy Mark gives us the powerful option “Click for three creds” once, and ALSO influences our draw function so that it’s worse in general but better when we’re poor. If we intend to draw when we’re poor, Easy Mark is great – it increases the utility of the draw function at a time when we’ll be drawing . If our deck doesn’t draw when we’re poor – maybe we have Codebusting and we’ll just do that when we’re poor – Easy Mark is hurting our draw function all the time, and in the times when it helps (we’re poor), it’s still not changing our draw function enough that we’re picking that option, so the cost is a lot greater. THAT is what the cost of “click to draw” actually means, and it’s not something you can quantify and “translate” into credits.

That’s a rough sketch of the Click Market, and a rough example of what card analysis using the idea looks like. I imagine as the idea gets more refined the analysis will also be. For now, I had a lot of words to get out and it left this whole thing kind of rambly. Sorry. I’m going to cut the article now – but expect a sequel shortly. I’m hoping that talking about this will help me focus my words on it.