Icing on the Cake

In my opinion, one of the hardest things to do when building your Corp deck is deciding which pieces of ICE you're going to use. There are so many of them, so many subtypes, and various icebreakers that your opponent might be using, and you have to consider all these things to build your defenses correctly.

I could post a list of all pieces of ICE in the game and tell you the server which you should place it, but I don't want to simply tell you "use this here and this there and GG". Not only it won’t be accurate, but I want you to understand why installing that particular piece of ICE in that specific server is the best move, so you'll be able to make your own future decision rather than wait for me to update the list.

John Nash and the ICE problem

You don't need to be familiar with game theory to understand this, but I'll need to go through this initial boring scientific part for everything to make sense. I’ll try to make it simple and practical, but I can’t promise you anything.

Game theory is based on an important axiom: people are rational. That means that I’m going to assume that the Runner will always make the optimum move he can make every click. This is not 100% true in real world, but – in high level competitions – I think it is safe to say that your opponents will act reasonably more often than not.

A move is defined as the optimum one based on its “payoffs” – which is basically a different word for reward. The Runner will do whatever rewards him or her the most. That means the only way to prevent the Runner from running on our servers is changing his payoffs - we have to make "running our server" a non-optimum move for the Runner. That's what ICEs do (ambushes do that too, but I won’t talk about ambushes today).

Why you don't need to protect your Archives at the first turn? Because the Runner won't get anything by running there, so it is not an optimum move even without a piece of ICE there. Why you do need to protect your R&D at the first turn? Because the Runner might get an agenda if he or she runs on R&D. The “might” is important here.

There will be situation where the Runner “might” steal an agenda in every game. R&D and HQ work that way, for instance. But the chances to find an agenda are different in different servers - the expected payoffs are different. I'll give you an example. Consider a Corp deck with a 45 cards deck, 3 assets, and 9 agenda cards that are worth 20 agenda point:

- Each agenda, on average, is worth 20/9 agenda points;
- Each card in R&D has a 9/45 chance to be an agenda, so a run on R&D has an expected payoff equals to 20/9 * 9/45 = 0.44 agenda points;
- Each card installed in a remote server has a 9/12 chance to be an agenda, so a run on a remove server has an expected payoff equals to 20/9 * 9/12 = 1.67 agenda points.

Obviously, these distributions of probabilities will change all the time over the course of the game. You’re not supposed to keep tracking them all during the entire game. The point here is to show you that, as different servers offer different expected payoffs, you won't need a perfect defense for every server, but to reduce the expected payoff until it is not worth it to run your servers.

Each thing in its proper place

You often hear people saying that you should place “economy pieces of ICE”, like Caduceus and Pop-Up Window, in your central servers because Runners run those servers more often. But is it the best explanation about why high level players do this?

Let's say I put a Shadow protecting my R&D. The Runner now has a new expected payoff, which involves me gaining 2 credits and the Runner taking a tag. While it was worth it to run R&D when there was no piece of ICE there, now it might be better for the Runner to spend clicks drawing cards until he find a program to deal with Shadow - that's what Icebreakers do, they improve the Runner's payoff. You can see where I'm going, right? You want to keep screwing the Runner's payoff so he or she won't run your servers because he or she will invest his or her time trying to improve his payoff to be worth to run again.

That makes sense for all pieces of ICE and all distributions of probability. Placing a Hadrian's Wall protecting the R&D and a Shadow protecting your main remote screws the Runner's expected payoff too, so it doesn’t really explains why placing Shadow protecting a central server is better than placing it protecting a remote server. But the thing is that different pieces of ICE affect the expected payoff differently.

The Shadow knows!

First, different subroutines change the expected payoff in different proportions. Consider the previous example, and let’s say I place a Wall of Static protecting my R&D. Until the Runner gets something to pass through my piece of ICE, the expected payoff of running on R&D was reduced from 0.44 agenda points to zero agenda points. But using it to protect a remote server would reduce from 1.67 agendas to zero agenda points, which is better for the Corp.

On the other hand, considering the Shadow drama again. Placing a Shadow protecting my remote would reduce the expected payoff from 1.67 agenda points to 1.67 agenda points, 2 credits for the Corp and 1 tag. Most of the time, this new expected payoff will still be better than the 0.44 agenda points from R&D, so my remote server would not be safe just because it has a Shadow in front of it.

Now you’re wondering why people use Shadow after all. Which brings us to my second point: different subtypes have different costs to be broken. Compare Garrote with Corroder, for instance. Garrote is far more expensive to install, making sentry pieces of ICE stronger, in the sense that icebreakers don’t solve the payoff issue as efficiently as they would if it was a barrier.

The true (or at least, the most accurate) reason about why placing economy pieces of ICE in a remote server is good is because we’re pairing a low expected payoff (0.44 agenda points) with hard to break subroutines. While using only “end the run” subroutines sounds good at the first moment, you must consider that the Runner will eventually have his or her icebreakers, which will severely hurt your ability to reduce the expected payoff from runs if all your pieces of ICE are Wall of Static.
(There’s also the surprise effect, like hitting an unexpected Neural Katana is better than an unexpected Ice Wall, but I won’t discuss that today)

Next Step

Now that you know how to protect your servers and how to create work compression, your next step is to learn how to bluff properly. Using ambushes is becoming more and more important as the game is becoming faster, and with a Jinteki deluxe expansion coming next month (according to FFG), there’s no better subject to talk about, right?

As always, feel free to ask anything. I understand that this article might be confusing to some people, but I really hope it helps. Seeya.

João “Hraklea” Almeida is a brazilian amateur card game player, the responsible for the Android: Netrunner league in Porto Alegre - RS -, in partnership with Lojas Jambô, and the writer of Root Cause, a bi-weekly series of articles about playing Anarch.