Concessions and Timed Wins

Netrunner General
62

I really don’t like the SoS because it does feel very random. I think what it really does is give veil of fairness to an inherently random structure. For example if you get paired up against a strong player in the first round and get swept you are at a significant disadvantage to making the cut because even if you claw your way back up you will do it through weaker opponents and thus not only have to make it to the points cutoff but rather exceed it because you will be lower in the SoS pecking order. Also small variations in the SoS based on players in the lower half that you played shouldn’t impact your chance of getting in. Just because you played the player that ended up 32nd and another player played the player that ended up 24th doesn’t mean jack but it could be the difference between getting in an not getting in.

Ultimately you have to incentivize players to want to get more wins. With the current structure there is minimal value to wins beyond what you need to make the cut. I like the randomization idea as long as it’s set up to put the top players in and then randomize the players with marginal points. Everyone would feel cheated if the top swiss player didn’t make it. But if you were right around where the cutoff would have been and don’t make it you could have always won more games and will feel less cheated. How often do you think if I sweep this last match I might make it? The RNG method is just an extension of that thought process and really changes the swiss mentality to let me get the most points in order to maximize my chance of making the cut. Which is exactly what we want players to be doing. It also never quite eliminates players and thus people in the middle still might feel in it and still play their best. Again, a very desirable outcome as far as player behavior.

I highly doubt this will ever be implemented but it is an interesting thought experiment on how to create a tournament structure that rewards players for playing good solid netrunner the whole time rather than playing the tournament structure. I’m more interested in player behavior than maximizing the probability of finding the best player. That is an interesting metric but less interesting to me.

1 Like
63

If you let a top N players get in guaranteed, you’re going halfsies on the idea. People will ID to get into the top N, just like they ID to get into the cut now.

The only exception is that you can award a player who is undefeated in Swiss a guaranteed place in the cut.

64

Fair enough. And we do want to highest scoring players in the last rounds to not just coast on in but rather still have matches that have meaning. We want people to have to fight for every win!

65

Even worse, of course, is losing (or even splitting), the first round to someone who isn’t terrifically strong, but is on the bubble themselves. For SoS purposes, it may not be the worst thing in the world to play your first round against Lucas Li, particularly if there are enough rounds remaining to reveal a more “accurate” prestige. But losing early to the player who comes in 30th in a 100 person tourney – which is probably in the reasonable realm of RNG outcomes for even exceptional players - might leave you in an unrecoverable position.

66

It’s nice that the author recognized the flaws of zero-point draws and I think he’s right that there could be a better system. I think he’s even further off base with banning ID/concession though.

1 Like
67

I agree the conclusion is wrong but like it’s an admirable thought process and investigation

68

RNG method is… ok, how can I put this? I don’t want to be rude, but to put it bluntly: It’s one of the worst possible solutions.

The whole point of tiebreakers is that the main criteria to determine the ‘best’ didn’t do well enough. So instead, your solution is to throw up your hands and just let people in randomly? You do understand how completely contrary to the purpose this is?

Sure, SoS isn’t perfect, but it’s a hell of a lot better indicator of player skill than random chance. The whole point is “Hm, these two players have the same Match points. But this one played Dan, Lucas Li, and Hollis. The other guy played three random scrubs…” In what logical capacity are you arguing that the guy that played people who didn’t win their matches is just as worthy as the guy who had to face people who won their matches more?

I’ve been on bubble often, and tiebreakers are annoying and only seem random. SoS is the best possible, after the unlikely occurrence of Head To Head. (Obviously if A beat B, then A should be ranked higher.)

5 Likes
69

Given that pairings in the first few rounds are heavily randomized, SoS as a tiebreaker is at least partially random as well. That doesn’t mean we have to throw it out (deep enough into any tiebreaker scheme, you’re going to have to turn to something that is either randomly determined or, at best, tangential to the regular course of gameplay like “total points scored”), but it’s not something that can be ignored. Someone who sweeps every match except for the one against (the undefeated) Player X is likely to have a significantly different SoS based on whether (s)he plays Player X in the first round, last round, or somewhere in between.

1 Like
70

So with these two different people that swept all their matches except against Player X, who faced stiffer opposition? The one who was swept in the first round, or the one who was swept in the last round?

71

You don’t get swept in the last round, because you ID. That reduction of data increases randomization in another way.

72

If true, then we don’t have even match points so we don’t need the tiebreaker to tell us that the person that went undefeated until final round ends the tournament ranked higher than someone that lost in round one and then didn’t lose another game.

Quick, did we need the higher ranked person to play their last round to tell us this fact, or did we already know it before he sat for his last round?

73

I understand this. What I’m trying to get at is a tournament structure that encourages players to win as many games as possible. The problem with the current structure is that there is little incentive to win all your games, just enough to make it into the cut. The idea isn’t just to let people in randomly. It’s to let them in randomly with probabilities based on how well they did. The main criteria to determine the best is full off lots of RNG based on the sample size anyways. That said, it is true that mostly the same players make it to the cut at big events.

I think a tiered cut of some kind is another idea worth exploring. Where players placing lower in the standings have an extra wildcard like round, say top 4 make it and then the next 8 have a 1 round playoff to make it into the cut. Or another idea is to start some players in the winners bracket and some players in the losers bracket. So they still made the cut but have to work harder and can’t lose.

Overall we want to give a greater incentive not to ID games by giving them value and not having a flat payout structure of making the cut or not.

3 Likes
74

The problem with trying to build a system where every game matters is that sooner or later you hit a point where they don’t anymore. If I sweep every game and everyone else has 4 prestige less than me then all I need to do is not get swept, so why wouldn’t I draw if I can? You can eliminate the cut, reduce it, add a random element, but there will always be a point where someone can math out that an ID will secure them a spot so they’ll do it. The only way to avoid that is to say that anyone who IDs gets eliminated from the cut bit then you go back to people doing it discreetly instead of blatantly so you don’t win even then. I too want every game to matter but at a certain point unless you go with something like “we put everyone who won a games name in a hat and now we will pick names until 4 people are picked” you can’t force people not to just draw when they are guaranteed a spot. (And nobody would like what I just proposed, because some regionals would be between 4 guys from the middle or bottom of the pack randomly and people who deserved their shot feel cheated)

2 Likes
75

You’re comparing the results of 4 rounds, and determining that in 4 rounds of play, player A did better than player B.

That’s less accurate than if there were 5 rounds of play, and in those 5 rounds player A did better than player B.

Here’s an example. Alice and Bob enter a 5 round Swiss. Alice and Bob each lose 300$ from their bank account every time they lose a game of Netrunner, so they don’t play this tournament for standings.

Round 1: Alice 2-0 Bob 0-2
Round 2: Alice 2-0 Bob 2-0
Round 3: Alice 2-0 Bob 2-0
Round 4: Alice 1-1 Bob 2-0
Round 5: Alice 0-2 Bob 2-0

Which player is better at Netrunner? Alice and Bob both fought hard to win every game to keep money in their bank accounts, and with determined efforts to win, Bob won a higher percentage of his games. Regardless of strength of schedule, it’s impossible not to conclude that Bob is the better player.

What happens when players aren’t highly motivated to win each individual game, and instead aim to make the cut, and can ID?

Round 1: Alice 2-0 Bob 0-2
Round 2: Alice 2-0 Bob 2-0
Round 3: Alice 2-0 Bob 2-0
Round 4: Alice 1-1 Bob 2-0
Round 5: Alice [ID]1-1[ID] Bob 2-0

Now Alice and Bob have equal prestige, from the exact same gameplay, just with Alice IDing the round she would have lost. What’s worse is that Alice almost certainly will have higher SoS because she started out winning and was paired with a stronger opponent in every round.

Using the knowledge that Alice is in a position to ID every time there, it’s not unreasonable at all to assign Bob a 25% chance of making the cut instead of Alice, based on the probability that Alice went 0-2 if she actually played that match out randomly. But if you manipulate the chances a bit more cleverly than that and instead just adjust them such that Alice actually plays out her game, you should get a cut that is at least as accurate as the one before, if not more accurate, and one where people actually play Netrunner and strike fewer backroom deals.

76

In all likelihood, the one who was swept in the last round. Since the stiffer competition was a result of a random pairing rather than the outcome of any particular decision by either player, however, I’m not sure why that would make it any less random.

77

Honestly you’re entering a prisoners’ dilemma style situation, where there are 3 outcomes, I make the cut and you might make the cut, you make the cut and I might make the cut, or we both most likely make the cut. Since an ID has to be from both parties, why would we each risk getting knocked out, when we can give ourselves a chance that isn’t dictated by another person? In the losing money example, if there some kind of prize worth losing 300 dollars per loss at the end, you can be sure I’m gonna ID to get a guaranteed spot in the cut, I’d rather take the chance of getting wiped out in the cut and lose there then try and save 300 dollars and maybe not even make the cut at all. The only way to make it around this is to completely cut out the cut at all and do only swiss, which puts you at the mercy of the pairing rng.

I just don’t see a solution that isn’t so random that people feel somewhat cheated.

1 Like
78

Pure Swiss is actually way more accurate than having a cut, mathematically and everything. The only RNG left is the unavoidable kind.

3 Likes
79

There is some incentive - the top ranked player entering the cut plays the lowest ranked, and gets to choose their side for the first match. That’s a non-trivial advantage in my experience.

3 Likes
80

Except that Alice shouldn’t ID in that scenario: She isn’t undefeated going into the final round. (Alternately, if less than cut*2 players are within 2 points of your current standing, you can ID.)

The only time you should logically ID is that if you ID, you get in on straight Match Points, without having to go to tiebreakers, since SoS is unpredictable. (Is it actually possible for Alice to have a worse SoS than Bob? If all of Alice’s opponents immediately bombed and didn’t win any games after facing her, and Bob’s did the opposite, can Bob have a higher SoS?) Just because SoS is more likely to favor the Nearly Undefeated player doesn’t mean it Definitely will.

Pretty sure I never gave a number of rounds. No matter how many rounds you play, if you go undefeated until the Nth round which is the final round, you’ve been facing higher opponents than the person who lost first round.

The reason we use the Swiss and Cut system is that we don’t have time to play Round Robin, and we also don’t want to keep one Loss from taking you out of the running for the top prize. We want to make sure that you earn your losses, and that it wasn’t just a fluke.

It may have appeared that I gave two contradictory viewpoints; first I said that SoS is more likely to favor the Nearly Undefeated player, but that it doesn’t Definitely favor them. Then I said that if you go undefeated until final round, you’ve been facing higher ranked opponents. This is why SoS is a tiebreaker, and not the main win condition. It is more random than Match Points and less random than Flipping a Coin. Match Points themselves are a product of some amount of luck, too. It’s just that they’re more influenced by skill than luck, as is SoS compared to a coinflip. (Or alphabetical. I’ve been in a tournament that cut equal match points by alphabetical last name.)

81

One loss doesn’t take you out of the running for top prize in a pure Swiss tournament either. After enough Swiss rounds, the undefeated player will get paired down. The common and more defensible argument for a cut is that it’s a better play experience and that it’s more exciting for players and spectators; I don’t think anyone actually argues that it selects the most skilled player the most accurately (an elimination cut is more clear about what number one victor it’s claiming to give you, it never goes to breakers like overly iterated Swiss can, but its outcome isn’t actually more accurate). Pure swiss iterated until there is a clear winner or SoS tiebroken winner is the most accurate thing you can use for sure. A cut is a way of damaging that pure accuracy for an exciting experience. It seems perfectly logical and consistent to me to damage that accuracy even more with graduated random entrance into the cut if you can further improve the player experience (actually playing the last round instead of performing a 50 minute handshake)

2 Likes