the way the corp scores agendas can reveal a whole bunch of information to the runner. Here is a situation I ran into yesterday. I’ve seen this sort of thing happen a few times in the past. My assertion: As long as you don’t do anything dumb, you’ve just won this game as the runner.
The Corporation :: NBN: Making News
Agenda Points: 0
Hand size: 3 (closed accounts, 2 unknown)
R&D Size: 35 cards remaining
Remote Server 1: Draco (rezzed, 0 str) -> bernice Mai (rezzed), Unrezzed card
HQ: Popup window (rezzed)
R&D: 1 Unrezzed Ice
The runner :: Kate
Agenda Points: 2 (Project Beale)
Programs: Ninja, Self-Modifying Code
Resources: Professional Contacts
Grip Size: 3 cards (Modded, R&D Interface, Daily Casts)
0. Mandatory draw
1. Play card into remote #1
2. Advance Card
3. Advance Card, rez SanSan, Score Project Beale (2 credits remaining)
This play has provided you as the runner with a wealth of valuable information.
1. What have you just learned?
2. What is your best course of action for this turn?
My feeling is you are now in a very strong position as the runner, to the point where if you don’t make any big mistakes, you default win this game.
Articles on Android: Netrunner strategy, as well as example situations for sharpening your wits on.
Archive for Geoff Hollis
- [+] Dice rolls
05 Aug 2013
Here's a game between JopeJope and I, with some new C&C decks. I talk a bit about some of the new tactics enabled by C&C cards, and berate myself over really bad plays.
YouTube Link: http://youtu.be/f_vXCnB5BpM
- [+] Dice rolls
05 Jun 2013
About a month ago, I did an analysis of the 40,000 game anonymized OCTGN dataset. Particularly, I was looking at what game balance was like for the most experienced players.
I took the 1% of players with the most amount of games played (called “the regulars”), and looked at their performance a) against the whole field, and b) against each other. There were a few take-home messages from this analysis.
1. The regulars have a high win rate against the field (duh). However, it is higher when they play as runner (68%) than when they play as corp (57%).
2. When regulars play against each other, there is a tendency for the runner to win (58% of the time).
3. Gabriel Santiago does seem to be influencing these results. When one of the regulars plays Gabriel, the runner wins 65% of the time. When looking at the other runners, the runner only wins 55% of the time. This is still reliably different from a 50/50 split between corp and runner wins, but it certainly is not in the 60-70% often-reported runner win rate for experienced players.
I was a little surprised with these analyses at the time. The runner was favored when regulars play against each other. The runner win rate (discounting Gabe) was lower than I expected, but the data seemed pretty clear on what they were saying.
One of my personal gripes with the analysis was that I was not entirely comfortable with the definition of “experienced player” I was working with. On average, people with more games played are going to be better than people with fewer games played. No disputing that. However, there is more to being good than simply playing lots of games; even within the group of regulars, there was a notable amount of variation in individual performance.
I’ve recently been playing around with the OCTGN league data, and while doing so it occurred to me that I do not have to rely on number of games played as a measure of player skill. The 40,000 game OCTGN dataset is ordered by time and has date stamp attached to each game. If I wanted to, I could calculate a more direct measure of player performance (e.g., Elo, Glicko).
So, I calculated everyone’s Elo and Glicko-2 ratings and redid my analyses from a month ago. A month ago, we looked at performance of “the regulars” -- that is, the 1% of players with the most number of games played.
This time, we are going to look at performance of “the professionals” -- the 1% of players with the highest Glicko-2 ratings. I’ve done the analyses with both Elo and Glicko-2 ratings, and the main results come out the same. However, I prefer to use Glicko-2 in this case, because it more conservative about who is performing extraordinarily well.
So, first, let’s look at the distribution of Elo and Glicko-2 ratings for all people who have played on OCTGN. The 99th percentile Elo rating is 1698, and the 99th percentile Glicko-2 rating is 1578. Nothing too odd seems to be going on with either distribution.
So how do the professionals perform? Well, they absolutely crush the field, winning 78% of their runner games and 72% of their corporation games. These are good players, as evidenced by their high Glicko-2 and Elo ratings. These are notably higher than the regulars’ win rates against the field (68% for runner, 58% for corp).
When the professionals play against each other, there is a decisive advantage for the runner (60.6% win rate over 407 games). This is a little higher than the 58% reported for the regulars, but the difference is nothing to make a fuss about.
However, it gets a little interesting when we look at the performance of individual identities. I’m only going to look at the seven Core Set identities, since there’s not really enough data points for the other identities represented in the 40k dataset (Chaos Theory, Whizzard, HB:Stronger Together, Weyland:Because We Built It).
Ok, a couple points I want to draw your attention to:
1. When the professionals play, all corporations are equally capable of winning. Jinteki is not at a notable loss, nor is HB notably better. HB shows what might end up being a slight edge with more data, but this effect is not statistically reliable within the current dataset.
2. Gabe’s win rate is 68% over 280 games. Noise is pretty good too, sitting at 62% over 276 games. Kate really seems to be lagging at 54% over 142 games.
I wanted to make this post for a couple reasons. First, the corp:runner imbalance might actually be a little bit more pronounced among the professionals than among the regulars. Gabe is certainly a culprit. Noise might be too.
Second, the fact that all corp identities seem to perform equally well when the professionals play is interesting. I would not be surprised if NBN and Jinteki had higher barriers to entry for playing well, but that does not mean they are bad (relative to the other corp identities). If you want to step up your game, it might be worth thinking about how you could actually make these difficult factions work well.
- [+] Dice rolls
flavortext introduction can be found over here
WARNING: Magnum Opus length post. Go get a snack.
I have a confession.
I have pretty much exclusively been playing Jinteki for the past three months. One of the reasons I haven’t put out a video recently is that I just don’t have any footage I want to show anyone. Call me greedy; I’ve been wanting to keep my tricks secret, at least until after regionals.
Gradius05 recently won the 26-person regionals in Plano, Texas, running a corp deck very similar to the one I’ve been working on for the past few months. In fact, I think he went undefeated as corp. I’ve really really been wanting to do some shop talk about Jinteki for awhile now, and I think I am finally going to break down and spill my guts.
I am going to discuss a Jinteki: Personal Evolution deck I have been refining for the past few months.
Ok, so first we need to make some observations about Jinteki. From a traditional perspective of what makes a good corporation, Jinteki is a hot mess.
1. From an economic perspective, Jinteki has terrible ice. Most is cost-efficient for the runner to deal with (e.g., wall of thorns vs. heimdall or tollbooth). The one exception is chum, but that has other problems -- particularly Yog-sucker problems.
2. From a security perspective, Jinteki has terrible ice. Wall of thorns is the only ice with a clear end-the-run on it but, as already established, Thorns has some other serious problems.
3. Jinteki has terrible in-faction economy. Akitaro and Dedicated Server do not give you any cashflow; you cannot directly use them to advance agendas. They are also both fragile and inconsistent.
4. Tricky and trappy cards gain value the longer the game goes on; longer time = more opportunities for combos to go off. Jinteki has lots of tricky and trappy cards. However, neither its economy nor ice support long games.
So, what’s a corp to do? You want better ice, right? So you import them from other factions. But then you also need economy to support those ice so you ... import it from other factions? There’s a serious problem here. Jinteki needs ice and economy from out of faction, but really doesn’t have the influence to take both.
So run neutral asset economy like PAD and Private Contracts! Well, the problem is that since your ice is so cheap to break, the runner is always going to have extra credits for trashing these things with impunity. If you defend these cards with ice, you will end up digging yourself into a deeper economic hole.
Runner economies are just really really good. HB:Core and Weyland may be able to win an economic pissing contest against runners sometimes, but Jinteki will almost certainly always lose a game if it comes to that. Magnum Opus is everywhere. Kati Jones too. Account Siphon is a gigantic kick in the face. The runner will force it to come to an economic war if she can.
But, Jinteki does have a very interesting tool. The best term I have for it is work compression. Jinteki can do a lot of little work here and there over many turns, each step requiring a single click and small credit investment, and then force the runner to match them click-for-click for that work, within a single turn or a) die, else b) let the corp score an agenda. I am talking about cards like Snare!, Data Mine, Ronin, Hosukai Grid, Neural EMP, False Lead, Nisei MKII, and Fetal AI. Project Junebug can be work compression if the runner runs on it. Edge of World reduces how much work compression you have to make the runner suffer, for you as the corp to capitalize.
Runners do have many tools to combat against work compression: Doppelganger, All-Nighter, Public Sympathy, Netshield, Joshua B, Diesel, and Quality Time all come to mind. However, few of these cards are runner essentials whereas cards like Magnum Opus and Kati Jones are.
I am going to be a jerk: many (most?) people I hear talking up Jinteki are inexperienced/naive/misinformed/delusional concerning Jinteki’s strengths, and the prowess of their jedi mind tricks.
In theory, Chum+unrezzed Neural Katana/Data Raven is amazing. In practice, you will never pull those combos off against a good runner. Stop thinking about cards in terms of their best-case potential; start thinking about cards in terms of their average-case potential.
In theory, running all 2-pt agendas makes sense; that’s more opportunity to flatline the runner, more sources of net damage, and extra chance of completely decking the runner. In practice, homogenizing your agenda point values takes away your ability to make the runner suffer from work compression. Decking the runner is a theoretical possibility, but not actually a realistic outcome.
In theory, getting a runner to run on a double-advanced junebug nets four cards. In practice, good runner decks are robust to losing specific cards; good runner decks can make use of anything, and rely on no specific card to function. In fact, double-advanced junebugs tend to be negative work compression (i.e., if the runner is getting 1 card/click and your only source of income is 1 credit/click, you are costing the runner 5 clicks at the expense of a 6 credit difference to not playing the junebug at all).
In theory, triple-advanced junebugs flatline a runner straight up. In practice, you are only getting the runner to run on triple-advanced junebugs in very edge cases. You cannot realistically treat it as a play in your playbook.
In theory, it would be cool if you could actually see a person’s soul by looking deep into their eyes, or read their darkest secrets off the twitches and ticks of their body. Sorry, you probably cannot. Unless you have had decades of formal, immersive experience with reading body language, you are no better than chance at inferring intentions from body language. If you believe otherwise, the better explanation is that you have a confirmation bias. I am not saying that inferring dispositions an intentions from behavior is impossible. There is a vast literature on this topic that says it is actually a very systematic process. But anyone who presents it as mysticism, some sort of ritualistic/prophetic skill, or The Key Ingredient To Success, is full of crap.
The variant of this is that Jinteki is all about mindgames. No. Again, the best explanation is that people are simply falling prey to confirmation bias. A:NR is a very structurally rich game. The vast majority of players will not share an appreciable amount of knowledge overlap with you. That knowledge overlap is required for mindgames to be effective. There are certainly exceptions, everyone pulls off the brilliant play every now and then, and extensive past history with a specific opponent changes things, but most people who say that mindgames are a big part about playing Jinteki well are simply being naive/unrealistically optimistic. Sorry.
I do not consider these serious points about Jinteki. However, I feel obligated to do lip service to the ideas since I see them surprisingly, and distressingly, often.
Defining Work Compression (hint: the most important point in this post)
Ok, getting back on track...
Jinteki’s most interesting tool is a form of work compression. If you would like a concrete definition of that, I am using work compression to mean “the ability to use clicks for actions across multiple turns, and then to force the runner to match you click-for-click within the space of a runner’s single turn”.
Economic bullying is a form of work compression, exemplified by Ash and/or deep servers. However, Jinteki usually sucks at that form of work compression. They do work compression through the runner’s hand size (in the case of Jinteki: PE), or more directly in terms of clicks (in the case of Jinteki:RP).
The corporation’s scored agenda points have an interesting interaction on work compression. Once the corporation gets to match-point, the runner needs to start being very aggressive about trying to fulfill every work-compression condition that might result in an agenda scored. If the runner is not, the corp can very easily exploit the fact and win.
Building and Playing my Jinteki Deck
So, this led me to a few ideas to keep in mind when building a Jinteki deck.
1. A good jinteki deck will need to be fast; the sooner you get to match-point, the more you can exploit the runner into taking work-compression risks. Note, this is completely counter to running all 2-pt agendas. You are going to need to run 3-pt agendas.
2. A good jinteki deck will not allow the runner to pay the costs of work compression using credits. You can always expect the runner to have a bank of credits since your ice is so terrible economically. Don’t fall for that trap.
3. Good play will not involve applying a steady stream of pressure on the runner’s work (whether that’s click work or handsize work). Good play will involve very “bursty” pressure; there are going to be lots of tempo swings within Jinteki games.
4. Some of your main tools for applying work compression on the runner (e.g., Snare!) require credits to activate. You will need a way to consistently float 6-8 credits if you expect to perform well.
With these things, in mind, here is my current Jinteki: Personal Evolution deck.
Deck Created with CardGameDB.com Android: Netrunner Deck Builder
Jinteki: Personal Evolution (Core)
Total Cards: (49)
False Lead (A Study in Static) x2
Priority Requisition (Core) x2
Braintrust (What Lies Ahead) x3
Fetal AI (Trace Amount) x3
Melange Mining Corp (Core) x2
Ronin (Future Proof) x3
Snare! (Core) x3
Pop-up Window (Cyber Exodus) x3 ■
Caduceus (What Lies Ahead) x2 ■■
Chum (Core) x3
Data Mine (Core) x3
Neural Katana (Core) x3
Enigma (Core) x2
Wall of Static (Core) x2
Rototurret (Core) x1 ■
Beanstalk Royalties (Core) x3 ■
Scorched Earth (Core) x1 ■■■■
Hedge Fund (Core) x3
Neural EMP (Core) x3
Hokusai Grid (Humanity's Shadow) x2
Total Agenda Points: 20
Influence Values Totals -
The Weyland Consortium: 11
Here is the starting gameplan for the deck.
1. Your economic ice (Popup, Caduceus) are preferentially going on R&D.
2. If you are up against a criminal runner, you want an ice in front of HQ as well. Preferably, Neural Katana on the first turn. You might lose an agenda but the loss of three cards will give you breathing room to score an agenda for yourself shortly afterward (chum+x is a very powerful remote server). If you are not up against criminal (or a known account siphon player), don’t even bother defending HQ. Let the implied threat of Snare! in an undefended HQ work to your advantage.
3. Your top priority is rushing to match point. Your ice is dead cheap. If you get lucky and score a chum in your opening hand, the practical cost of your remote server is 1 credit until the runner’s rig is assembled. If money is tight, you might opt not to rez non-economic ice protecting central servers.
4. Once you are on match point, your gameplan may change. If the runner’s rig still is not complete, you are balls-to-the-wall rushing out a last agenda. If the runner’s rig is complete, you are instead trying to orchestrate a situation where you can put the runner under a massive amount of work compression in a single turn. This is going to involve some very meticulous setup with data mines, Fetal AIs, and maybe a timely hit of Snare! on R&D. At this point, the runner is usually panicking and trying to control R&D. Snares (and Fetals) in R&D work to your advantage. You might start defending R&D with a few ice, but more importantly you want to be consistently floating 6-8 credits so you can capitalize on opportunities to apply work compression.
Lessons learned through trial and error
Do NOT try to bait the runner into running a Snare! facedown in a remote server if you cannot follow that up with a kill.
DO lay down Braintrust, Fetal AI, and Snare! undefended when you are on match-point, are sitting on 8+ credits, and the runner only has 4 cards in hand.
Do NOT play-advance-advance Fetal AI unless you are on match point, or have no other agendas in hand and the runner gives you an opening for an uncontested score.
DO consider playing fetal AI and advancing it once if the runner has 4 or fewer cards in hand, you have hosukai grid, Priority Requisition sitting in hand, 8+ credits, plus 2 data mines and/or two neural EMPs.
Do NOT play Ronin early. Your priority is scoring agendas, not killing the runner. Killing the runner is the consequence of work compression, and that is something best applied when you are on match point.
DO play Ronin facedown in a server when the runner is at 3-4 cards in hand and you still have 4+ Snares/Fetal AIs in R&D and a couple neural EMPs in hand. Next turn, proceed to advance it three times and leave it. start building a new server afterward.
Do NOT play single cards in unprotected remote servers.
DO play three cards in separate unprotected remote servers if 1) there is a permutation of cards that would threaten the runner’s life by hitting two of them in a turn, 2) at least one of those cards is actually part of that permutation, and 3) you can get value out of the runner not running at least one of those cards.
Do NOT late-game bluff with priority requisition when the runner has a full grip.
DO late-game bluff with priority requisition always when the runner is sitting on 3 or fewer cards in hand. For that matter, do this with any advancable card.
So, how does this deck perform? I have been having ridiculous success with it. Db0 recently released the 40,000 game anonymized OCTGN dataset. I managed to figure out which user ID corresponded to me (not that hard -- there is only one person who has been obsessively playing Jinteki: PE for the past three months), and calculated my win rate.
I filtered out my first 30 games with the identity; they were largely trial-error learning and I was making lots of poor plays. That still left me with something like 130ish games to look at. My overall win rate with the deck was 74%. Of those 130ish games, my win rate against experienced players was about 65% over 30ish games. For comparison, my win rate with HB against the same group of experienced players is only about 45%.
So is this deck the holy grail? No.
It is a very strong deck. However, it has some serious shortcomings.
Public Sympathy singlehandedly shuts this deck down. So does Net Shield. Net Shield still isn’t seeing alot of play, but Public Sympathy is getting table time. It makes the job of applying work compression on the runner’s hand incredibly difficult.
Crypsis can be an annoyance. He prevents the net damage of data mine, but he doesn’t always prevent the work compression of data mine since you have to load crypsis with a virus counter first. The trick is to catch Crypsis when he has no virus counters loaded on him.
Darwin, on the other hand, is a serious issue. Darwin will probably be seeing lots of of playtime in the next couple months. He’s interesting and fun to tinker with.
This deck may not pass the Gabe test. Although my winrate is about 65% against non-Gabing experienced players, it drops to about 45% when only looking at Gabe games. Don’t get me wrong -- that’s still pretty good. It’s just not stellar.
This deck presents many opportunities for the corporation to make critical mistakes. After playing about 200 games with it, I still make a serious blunder every 5-6 games. I did not really begin having success with this deck until I played it about 30 times. Even now, I am constantly refining my play with it.
To play this deck well, you need to constantly track the number of cards in the corporation’s hand. This can prove to be a hinderance in face-to-face play (asking a person their hand size can give away your intentions if you are not careful).
This deck suffers severely from the Corporation Repeat Play Effect. As such, I have a few variants of it that substitute the ronins for Junebugs/Edge of World. Occasionally, I will also splash in an Archer. I almost always regret the Archer.
The singleton scorched earth may seem out of place; I assure you it has pulled its weight for a long period of time. About a third of my wins against Gabe involve that single scorched earth, and about a fifth of my wins against other runners involve its presence. That said, Kati Jones + resource-heavy Andromeda decks really diminish its value against the range of runner decks you might see. I am definitely thinking about taking it out.
I am well past a 3,000 word count. This seems like a good place to stop.
- [+] Dice rolls
09 May 2013
A little bit of background: Susuexp is currently conducting an ongoing analysis of the 40,000 game anonymized OCTGN dataset. If you’re a stats nerd, it is definitely a must-watch. The level of detail and types of analyses are fantastic and provide much to think about.
In Susuexp’s most recent episode, The Community, he presented a series of graphs towards the end that suggested for very frequent players, the corporation and runner sides were more or less balanced, and the corporation may even have a slight edge. However, it was commented that the results needed to be taken with a grain of salt; there were not enough data to draw strong conclusions.
For at least two reasons, I was very skeptical of these particular graphs Susuexp presented. I am definitely one of those “very frequent players” on OCTGN. My subjective perception is that, when playing against other regulars, the runner most certainly has the edge -- probably in the area of a 70% win rate. Second, I would have to guess I have played at least 100 games against other regulars in the past couple months, just by myself. Given that other regulars must be playing against each other as well, I have a hard time believing that the data set contains less than 500 games played between OCTGN regulars. 500 datapoints is quite sufficient for answering any sort of simple question like “which side is favored?”
So I decided to do some followup work on the 40,000 game dataset. If you’ve been following Susuexp’s video series, you’ll have noticed he’s doing some fairly complicated time- and order- dependent analyses. This is a good thing -- someone needs to be doing this stuff. However, the (very granular) types of analyses he is performing spread the data a little thin when you want to ask questions like, “what is game balance like for the most experienced players?”
So with that in mind, I am just going to do some very simple plots of corp and runner win rates in games involving “the regulars”.
The 40,000 game dataset contains data from 3858 unique players. Most have only played a single game or two. A couple have played a thousand or more. The mean number of games played by a single player is 22. The median is 4. The 90th percentile is 58 games played. The 95th percentile is 100 games played. The 99th percentile is 250 games played.
I am curious about “what does game balance look like for the most experienced players?” We will call this group “the regulars”, and arbitrarily define the group as everyone in the 99th percentile for games played.
In addition, we are only going to be looking at games that are, at minimum, the regular’s 200th game -- this is simply to make sure we are only looking at games where the regulars are at their “most experienced”. So, say a regular has 250 games under their belt. I am only looking at their 250 - 200 = 50 last games played.
We are going to walk through a few questions. First, let’s ask the question “how do the regulars perform across the field of all players?”. The most important graph is the rightmost one; it contains all the games from all the regulars, combined.
The regulars perform quite well against the field. Their win rate for both corporation and runner is above 50%. However, their win rate as runner is higher (about 68%) than their win rate as corporation (about 57%). The regulars are strong players, and win far more than their fair share of games. This shouldn’t be surprising; these are the players with the most experience playing Android: Netrunner. They do have an easier time winning as runner than corporation, though, which is interesting.
All of the smaller graphs to the left are data for the ten individual most frequent players. Note that the patterns of “the regulars are better than average” and “the regulars win more often as runner than corporation” seem to hold at an individual level.
Ok, let’s move onto a new question: “how do the regulars perform when playing against each other?”. Now we are only going to look at games played between two regulars. Additionally, a game will only be included for analysis if it is both player’s 200th+ game. This leaves us with a very healthy n = 532 games. Again, the most important graph is the rightmost one; it contains all the games from all the regulars, combined.
Corporation win rate is 42%, which means runner win rate is 58%. This is reliably different from a 50/50 split (n = 532; p < 0.001). Just to make sure this difference is not due to “a couple of regulars who are very weird players”, I have again plotted the individual win rates for the ten regulars with the most number of games under their belt. In fact, there does seem to be variability from player to player, and at least three of these players do not follow the overall trend (#3837, #2100, #3088). Note, though, the n’s are fairly low for individual player games.
Taken as a whole group, the runner wins more often than the corporation when the regulars play against each other.
But... but...! Maybe this is just because of that dastardly Gabriel Santiago! Yeah, ok, fair. Let’s ask a third question: “how do the regulars perform when playing against each other, and neither is playing Gabriel Santiago?”
When excluding games involving Gabriel Santiago, we are left with n=368 games. Corporation win rates go up (to 45%), runner win rates go down (to 55%). This is still a marginally reliable effect (two-sided p = 0.08; one-sided p = 0.04). Gabriel is certainly influencing the data, but he alone is not the entire story. There does seem to be a bias in favor of the runner for the regulars. However, contrary to my expectations, it is not in the 60-70% range.
So, where does the “runners will win 60-70% of the time between two experienced players” statement come from? It is certainly not supported by these data, which actually pin the runner win rate somewhere between 55% - 60%.
I think there are two possibilities.
1. Go back to the first set of graphs. When you look at the regulars’ win rate against the entire field of players, their runner win rates are in the 70% range. Perhaps experienced players are having a hard time mentally separating their games against other experienced players from games against everyone else, and are actually providing estimates that are contaminated by their “win rates against everyone”. This would be a very reasonable, non-surprising error for humans to make.
2. Go back to the second set of graphs. Although the overall runner win rate between experienced players is 58%, there are in fact some regulars in the 60+ range. Notably, these also tend to be players with less than 50% win rates as corporation against other regulars. It could just be these are the people who are loudest on the issue.
So, in conclusion: I don’t think it matters how you cut it. Between experienced players, there is a tendency for runners to have the edge. There are certainly exceptions, but they are just that: exceptions. The runner’s favor is certainly not in the 60-70% range, and Gabriel Santiago is certainly influencing the data (to an extent), but at this point I would be very surprised to see someone find a reasonable way to argue to the contrary; the runner is currently favored when regulars play.
- [+] Dice rolls
I recently had the opportunity to discuss Android: Netrunner with Lukas Litzsinger. So we're going to do something a little different this month. Instead of a strategy article or video, we're going to hear Lukas's thoughts on the design of Android: Netrunner, the current metagame and balance, and ways that Android: Netrunner differs from the original Netrunner.
I found it particularly interesting to hear Lukas's take on faction and identity balance, as well as tournament play. There are definitely a bunch more questions I wanted to ask him, but I really did not want to hog his entire weekend. As it stands, I am quite happy with the insight he has provided.
Introduction .................................. 0:00
Deckbuilding .................................. 0:58
Corporation Economy Cards ..................... 4:12
Identity Balance .............................. 6:40
Faction Balance ............................... 10:20
The Tournament Structure ...................... 13:20
The Trace Mechanic ............................ 16:50
Out of Faction Influence ...................... 19:10
Identifying With Specific Factions ............ 20:50
Creation and Control .......................... 24:20
Fast Advance .................................. 26:30
Outroduction .................................. 30:30
You can download the interview here.
- [+] Dice rolls
- [+] Dice rolls
Apologies. This post is of epic length. I hope you agree that, despite its length, every point was important to make.
George and Helga play A:NR together. Helga favors a Noise virus mill deck and George favors an HB fast-advance deck. Historically, when Helga and George face these decks off against each other, Helga wins 80% of the time. When Helga wins (80%), George consistently scores at least 4 agenda points. When George wins (20%), Helga scores 0-6 agenda points, with a uniform distribution.
On the flipside, George favors a hyper-aggressive Gabe deck. Helga’s favorite corporation deck is an NBN trace deck. Historically, when Helga and George face these decks off against each other, George wins 65% of the time. However, when George wins (65%), is it almost always a complete blowout (Helga never scores more than 3 agenda points, and the vast majority of the time scores 0 or 1). When Helga wins (35%), George scores 0-6 agenda points, with a uniform distribution.
If George and Helga faced off in the Swiss rounds of a tournament, who is expected to win the match? Remember: Helga’s win rates are 80% as runner and 35% as corporation. George’s win rates are 65% as runner, and 20% as corporation. Based only on these numbers, we should expect that Helga is clearly favored to win the matchup. However, what might surprise you (and certainly surprised me), is that George is actually favored to win this match.
George’s prestige earnings:
65% runner win * 20% corp win = 6 prestige 13% of the time
65% runner win * 80% corp loss = 4 prestige 52% of the time
35% runner loss * 80% corp loss = 0 prestige 28% of the time
35% runner loss * 20% corp win * 40% match win = 4 prestige 2.8% of the time
35% runner loss * 20% corp win * 40% match loss = 2 prestige 2.8% of the time
35% runner loss * 20% corp win * 20% match tie = 3 prestige 1.4% of the time
George’s expected prestige points: 3.07
George's chance of winning the match: 65%
Conversely, we can figure out that Helga is expected to earn 6.00 - 3.07 = 2.93 prestige points on average.
Here is a more straightforward case: Effe has the king of all runner decks. He has literally never lost, and proudly sports a 100% win rate. However, the corporation regularly scores 4-6 agenda points against him. Frank’s runner deck, on the other hand, wins 80% of the time against all corporation decks, but when he does, it is a complete blow-out (corporation earns 0 prestige points). If Effe and Frank face off in a match, who is favored to win? Again, the answer may be surprising: Frank is favored to win this matchup.
Frank’s prestige earnings:
80% runner win * 100% corp loss = 4 prestige 80% of the time
20% runner loss * 100% corp loss = 0 prestige 20% of the time
Frank’s expected prestige points: 3.2
Frank's chance of winning the match: 80%
This is very interesting. Having a strictly better win rate does not necessarily translate to earning more prestige points, nor to having a better chance of winning the match.
I do not presume to know why prestige points are awarded for a match win, nor am I privy to the reasoning that went behind assigning 2 prestige points to a match win over some other value, or why match points were chosen as the primary tie-breaker as opposed to something else.
What is clear, however, is that the current tournament framework rewards blowout wins. In fact, you can reasonably sacrifice your overall win rate as long as you can can find ways to ensure the fewer wins you do have are more likely to be blowouts. By doing this, you come out ahead in terms of prestige points AND match wins. This is especially important in elimination rounds.
What does this mean for tournament deck design?
If you are playing a tournament, have ambitions of winning, and actually have the relative skill to put yourself in contention to win, these observations prescribe a very specific deckbuilding philosophy: if one of your decks is favored to win and the other is not, optimize your winning deck to win by a landslide, even if this means the deck might lose a little more overall. Optimize your losing deck to score at least some agenda points, even if it means losing more overall. In short: play an aggressive criminal deck, and HB fast-advance. Be very wary about sacrificing agendas (e.g., archer). Cards like this might increase your winning chances overall, but when you lose, it makes the loss more likely to be a blowout. Be very wary about playing late-game control decks on the runner side (e.g., big-rig kate + medium, noise mill) that have a difficult time busting early remote servers. Even if your deck has an astounding win rate, you’re putting yourself in a situation where you probably won’t win the match after the table turns and you’re up against a hyper-aggressive runner that doesn’t win games as often, but wins decisively.
Results from the latest OCTGN tourney make it very clear that, for strong players, runner is overwhelmingly favored to win the match. You can leverage this observation into more match points by playing hyper-aggressive runner decks, and corporation decks that aim to reliably score *some* points at the cost of actually winning.
What does this mean for tournaments?
I hope you are all alarmed by this advice. It is meant to be tongue in cheek. There is a problem with A:NR’s underlying tournament scoring. I am explaining how to exploit that underlying problem, rather than figuring out ways to fix it. The real purpose of this article is to discuss A:NR’s tournament framework, identify its problems, goals, and try to make progress on improving the system overall.
It is worth pointing out, I am starting from a basic assumption that better win percentage should necessarily translate to better placement in tournaments in the long run. If you do not share this basic assumption, we are probably at an impasse. That’s fine. I can respect that my opinions are not shared by everyone. You may be satisfied with the current tournament framework. I am not.
An important part of any project is identifying what your design goals are: what is the purpose or function of the thing I am about to create? I have found little information on FFG’s design goals for tournament play. I understand they want to foster a tournament scene, but I have no clue why. Is a tournament scene a means to the end of increasing awareness and hype about the game? Is a tournament scene a mechanism to foster structured, competitive play? Are tournaments meant to offer a form of ranking/evaluation? Or are the primary purpose of tournaments to foster social cohesion and community development? Depending on what the design goals are, the current tournament format may or may not be sufficient.
The closest thing to a design goal I have read is in the official tournament rules, page 2: “Please remember that these tournaments are designed for players who want to celebrate and enjoy the most challenging aspects of Android: Netrunner.” I interpret this to mean that the primary design goal of tournaments is to foster structured, high-level, competitive play. Hopefully it is clear that the current tournament scheme does not satisfy this goal, because of the disconnect between game win rates vs. prestige points earned / winning the match.
The issue of Game wins, Match wins, and Tournament wins
My current belief is that all issues with A:NR tournaments stem from the fact that goals of winning an individual game, the overall match, and the tournament are divergent. As has been discussed ad nauseam, putting yourself in a position to more likely win an individual match and putting yourself in a position to more likely win the tournament are sometimes conflicting; there are times where intentional draws in the last rounds of Swiss are the best actions for winning a tournament.
Less often discussed, however, is the fact that putting yourself in a position to win an individual game sometimes conflicts with your ability to win the match. AndrewRogue has discussed this in his blog. I have provided examples (above) where players can actually win individual games less frequently, but as a result put themselves in a better place to win the match (and tournament) because of it.
Correct play for winning individual games is not correct play for winning individual matches. Correct play for winning individual matches is not correct play for winning individual tournaments. This creates uncomfortable decisions for players. Furthermore it means that “A:NR skill” does not directly transfer to “tournament placement”. Good tournament play is a skill unto itself, which will create a rift (maybe just a small one) between general A:NR play and tournament play. If FFG’s goal for tournaments is to “celebrate and enjoy the most challenging aspects of Android: Netrunner”, there is a problem here.
FFG has implied in the official tournament rules that players should be playing with the goal of winning individual matches (see: Unsportsmanlike Conduct, p.2 of official tournament rules). Decisions that increase your chance of winning a tournament, at the expense of winning an individual match, are unsportsmanlike conduct. In reality, this is not a big deal; it effectively amounts to saying “intentional draws are not allowed”. You may or may not have objection to this depending on your personal philosophy, but the fact is this is an easy rule to follow once established.
However, we have to assume that a decision to increase your chance of winning an individual game, at the expense of winning an individual match, is also unsportsmanlike conduct.
So suppose we’re in a situation where Anne and Bob are playing together. It is the sixth round of Swiss. Anne has 16 prestige points and Bob has 18. It is very clear that the cut to final elimination will be 21 or 22 prestige points. Anne managed to win the first game (she played runner) 7-0. Second game she is playing NBN and, in her opening hand, draws breaking news. She is guaranteed a match win by scoring breaking news from hand. However, if she does this, she is almost certainly throwing her chances to win the game. To progress in the tournament, Anne NEEDS to play for a game win (it is the only way she can score 22 prestige points over all). However, in the strictest sense of the rules, she is engaging in unsportsmanlike conduct if she plays for the game/tournament win rather than the match win.
I am pretty sure no tournament organizer in their right mind would expel Anne for unsportsmanlike conduct if she did not score breaking news from hand on turn one. The fact is, though, it would be unsportsmanlike play by the letter of the rulebook. So if we allow Anne to play for the game/tournament win here, rather than the match win, can we frown on people playing for a tournament win rather than a match win by taking an intentional draw? Perhaps some people can rationalize such a distinction. I cannot; I find it completely arbitrary that a “match” is somehow a more meaningful unit of measurement than a “game” or a “tournament”. Ultimately, I think correct tournament etiquette hinges on subjective philosophical belief. This presents opportunity for conflict when an event necessarily brings together people with different subjective philosophical beliefs. Enjoyment and mode of interaction should be a byproduct of game design, not the philosophical beliefs you hold entering the event.
That said, I respect that if a tournament has expressly forbidden intentional draws, or has expressly forbidden playing for a game win, I should follow those rules. I may philosophically disagree with those rules, but through the act of signing up for a tournament with those rules in place, I agree to indulge a set of philosophical beliefs that I may not share -- at least for the duration of the event. I believe the appropriate saying is, “my house, my rules”.
What would be nice, however, is if tournaments were structured in such a way that philosophical or moral disagreements would not arise in the first place. That is to say: we should strive for a system where game wins, match wins, and tournament wins are not at odds with each other.
Reconciling Game wins, Match wins, and Tournament wins
Now that all the theory and premise is covered, let’s get down to business: how can we feasibly reconcile the game/match/tournament win issue?
Option 1: eliminate prestige points for a “match win”. Instead, a match consists of 2 games (one on each side). Each game you win, you earn 2 points. That’s all. So you can either earn 0, 2, or 4 points. If the second game goes to time call it a draw, in which case the score will be 3-1.
Correct play for game wins is now consistent with correct play for match wins. If we revisit the examples of George vs. Helga and Effe vs. Frank at the very beginning of this article, there is a pleasant change:
George’s expected prestige points are 1.7, and Helga’s are 2.3 (previously 3.07 and 2.93).
Effe’s expected prestige points are 2.6, and Frank’s are 1.4 (previously 2.8 and 3.2).
Also if we revisit the example of Anne vs. Bob, these weird situations where you should sacrifice a game win for a match win does not arise. Increasing your chances of winning an individual game will always increase your overall placing in the tournament.
However, there are two problems. First, intentional draws still serve tactical purpose; there are times where ensuring a 2-2 draw is better than risking a 0-4 loss if you want to make the cut to elimination rounds. Second, such a system would result in a huge number of ties (appx. 50% of matches would end in a tie at top level play). I cannot really evaluate if this is bad from a design standpoint. However, it is certainly aesthetically displeasing.
I think eliminating prestige points for a match win is a strict improvement over the current system, but that doesn’t mean it is satisfactory.
Option 2: eliminate the concept of a match entirely. Each round you play one game, but the side you are on alternates from round to round. Apparently competitive chess does this, and there are known pairing algorithms to make such a system work (more or less).
The pairing algorithm used is not perfect, and players may be paired the same color on two subsequent rounds. This means that, occasionally, you might play 6 rounds -- 4 as black, 2 as white. There is a small cost for unluckily paired colors in chess; White is slightly favored (52-55% chance of winning). The system isn’t perfect, but it is workable.
The problem would be a little worse for A:NR. At high-level play, the runner is favored to win about 60-70% of games. Getting bad colors in a chess tournament sucks, but isn’t overly debilitating. In A:NR, the relative strengths of each side are quite disparate. Getting 2 runner pairings and 4 corp pairings is a large handicap over someone with 3/3 or 4/2 pairings. Also, intentional draws (playing for the tournament win rather than the game win) are still a valid strategy.
Option 3: Increase the threshold for the finals cut. Byronczimmer noted that in the OCTGN tourney 2, we cut to the top 16 out of two fields of 64. If a player were to win all six of their games 4-2, (24 prestige earned total) they would have made the cut to the top 16 in both divisions (21 prestige threshold). If, however, we made a cut to the top 8, a player that went 4-2 in all six rounds would not have made the cut to the top 8 (25 prestige threshold).
Trying to play for blowout wins as the runner and minimizing the margin of loss as corporation become less effective strategies in the swiss rounds. Correct game-winning play is brought in line with correct tournament-winning play. Correct match-winning play is still in conflict.
Also, if you do manage to get into the elimination rounds with decks like George above is playing, you will probably have the lion’s share of games, since a tighter cutoff threshold necessarily selects for decks that consistently win, but sacrifice the potential to earn prestige for the match win in the process (if you can’t wrap your head around this statement, please see the two examples at the beginning of this article).
Tighter cutoff thresholds reduce the frequency that goals of game/match/tournament wins are in conflict, but still does not address the underlying problem.
Option 4: Make it so events have two simultaneous tournament tracks: a runner tournament, and a corporation tournament. Goals of game wins and match wins are now necessarily aligned. Intentional draws may still serve a tactical purpose for winning a tournament.
In a world with unlimited time constraints, I think this would probably be the ideal solution. However, it is not clear (to me) that a pairing algorithm could be devised that would not result in massive scheduling conflicts.
Unfortunately, I do not have a definitive solution to the problem at hand. I think that Option 1 currently presents itself as the most promising solution of the bunch, with the addendum that intentional draws are left to the discretion of the TO or some larger governing body (FFG). I find it incredibly dissatisfying that the current tournament structure sets up situations where people with worse win rates plausibly perform better than people with better win rates. That is simply unsatisfactory in the context of fostering competitive play.
A better tournament structure needs to be devised. I hope FFG agrees with this sentiment. I also hope we can collaborate as a community to design a better system.
- [+] Dice rolls
I've managed to put together a dataset for the games/matches from the second OCTGN tourney. You can find it here. There are still a couple matches missing from the records. I have also filtered out games involving players that dropped, or agreed to intentional draws. However, that still leaves 500+ games played, split between the US and European divisions.
I have done a preliminary analysis on the data. General OCTGN data have suggested that the corporation and runner are evenly matched. However, as I noted after the first tournament, it actually seems as if runners have the advantage when advanced players play, and corporations have the advantage when novices play.
With swiss rounds of the second tournament complete, we now have another opportunity to look at how game balance interacts with player experience. Below is a graph plotting player skill against chance of winning as the corporation. There is a very clear, negative correlation: as the overall skill of two players increases, the chance of the corporation winning decreases. Player skill within a match is simply the sum of both player's earned prestige points throughout the tournament.
This is an exact replication from what we saw last tournament, but this time with a larger sample (128 players vs 64 players). This should close off any debate as to which side has the advantage: the corporation has the advantage between novices. As players get more experienced, the runner progressively gains the advantage.
An interesting question about tournament play is, given the option, which side should you play first: runner or corporation? Conventional wisdom suggests you should play the runner first, as long as you're not going to be pressed for time.
While I completely agree with the reasoning laid out by AndrewRogue in his article, I am inherently distrustful of theory -- no matter how sound it seems -- without empirical evidence; real-world systems often behave much differently than we intuitively think they do. Tourney data provides us with a good opportunity to actually test such theories. So, here are the data (again, broken down by player skill):
interesting. Yes, against two very good players, playing the runner first carries a distinct advantage. However, against two novices, playing the corporation first carries a distinct advantage. Combined with what we already know about the corp dominating novice games, and the runner dominating advanced games, perhaps the best advice is: if given the option, first play the side that has the advantage.
I also want to draw attention to the size of this advantage. If we take only data from the top half of matches (i.e., matches where the summed prestige of both players is higher than the median), the player who runs first will win the match 63% of the time. When you look at this number as a percentage, that doesn't seem like a big number. However, consider this as an odds ratio. A 63% chance of winning is 1.7:1 odds -- you will win 1.7 matches for every 1 match your opponent wins. That is a HUGE advantage for being on the lucky side of a coin flip. This advantage compromises A:NR's ability to have a serious tournament scene.
Alright, so there is my preliminary analysis. I will do a little more next week, but I think these are two very meaty findings and probably warrant pondering. I am becoming increasingly disillusioned with A:NR as a tournament game (I still like the game, but just not as a tournament game). I will post a lengthier theory article on my thoughts soon.
- [+] Dice rolls
13 Mar 2013
- [+] Dice rolls