Isildur1 and the week that changed online poker | Hacker News Reader

Isildur1 and the week that changed online poker

blogs.cardrunners.com

87 pointsjskopek59 comments

I don't play cards. Without a doubt, I loved this story and description.

"Durrrr can be dumb. Sometimes he makes mistakes, he tilts, he makes clearly –EV calls and he sticks huge stacks in with rags against the nuts over and over again. Durrrr is often reckless, sometimes emotional, and even at times irrational. Durrrr is fallible. He is imperfect. And yet, somehow he wins. He outplays, he outmaneuvers, and outthinks. He reached the top. He beat everyone. He became the king. He symbolizes the human in all of us, and he bears testament that one does not need to be perfect, unphaseable, untiltable in order to become great"

Am I the only person who found that article so jargon rich to be almost unreadable? I wonder if people who hear me blathering on about IPsec Transport sessions and RFC 2893 tunnels have the same experience.

I translated some of the jargon:

NLHE: no limit hold'em x/y--size of the blinds, the ante you pay before seeing the cards in hold'em.

grinders--someone who plays a lot of poker and makes a little bit of money. to make a lot of money you have to "grind" out a bunch of wins over a long period of time.

the nuts--best hand possible to win a hand of poker buyin--amount of $ you are allowed to start a game at a table with.

PLO--pot limit omaha, a kind of poker

EV--expected value

2p2--popular poker forum Two Plus Two

There are poker slang glossaries online available for further enlightenment.

It was written for the poker community, not a general audience. As with any complex field, you need a specialized vocabulary to communicate effectively.

Further to the other reply:

- the rail/railbirds - observers to an online poker match

- hyperaggro - a very aggressive player (bets and raises a lot), such as Isildur

- n-tabling - playing multiple online poker games simultaneously.

- HU - heads-up, a two player poker game, such as the one between Isildur and Tom "Durrr" Dwan

- degen - dengenerate, someone who takes too many risks and plays at stakes higher than they can afford

Rereading the article, I can see what you mean. From the outside it can look like all the jargon is just to exclude outsiders, but it really is needed to communicate effectively.

Ah, this article makes me want to go back to poker...

It sounded to me like he was baiting them, hustling like a pool shark. He'd make a win, then lose most of it to give the apparently superior player the taste of success and make them risk the big money then he took them for it. Rinse, repeat.

The bit where the writer said Isildur went "brb" and came back and won from then made me think he brought some other force into play, either a companion, a hack or a bot.

It's kind of hard to bait and switch repeatedly over the number of hands this guy was playing. I think the number 10,000 hands was floating around in the article somewhere, for one match alone. And at the highest stakes players' edges are pretty thin. It's foolish to think that

a) you have a significant and consistent edge over most other high stakes players (Tom Dwan, Brian Townsend, Patrick Antonius, etc.),

b) those players would be fooled repeatedly by bait and switching without adjusting their game, and

c) you could run such a strategy successfully without having ever played those players at those stakes online before.

More likely he just took a break to calm himself and get his mind right.

Poker is intensely psychologically draining and the biggest enemy of winning poker players is tilt. http://en.wikipedia.org/wiki/Tilt_%28poker%29

Most good players will constantly asses their play for signs of tilt and when they spot it, stop and take steps to address it before continuing.

Learning to recognize when I'm going on tilt has been one of the most important processes in my becoming a better poker player. I found that underplays threw me off my game far more than anything else, so I've worked hard to take my chances when I get them, and quickly move past it when I don't. It wasn't until I lost about $10 (mind you, my bankroll was $20) after getting nothing out of a straight flush that I realized how much such things affected me, and took steps to address it.

Obviously there's a huge difference between $10 and a few hundred thousand dollars, but the same core concepts apply. It's difficult for people who haven't played a decent bit of poker (even at very low stakes) to understand how psychologically draining it can be, and how much the player's frame of mind affects the game.

I highly doubt this. Have you ever hustled? It's extremely hard to hustle the hustlers. He was playing the top players in the entire world, not some fish filled pond in Atlantic City.

I'm responding to teh report of how things went down, no I haven't ever hustled, have you? When I tasted online poker I could only ever end about 3rd hand in the competition when top 2 were winners; never played for money online.

It's next to impossible to hustle someone in poker regardless. You can only misrepresent your skill for a very short period of time before the losses become too great to overcome. In most cases you'll never be able to setup a huge bet to recover those losses. (Especially with capped buy-ins of 100-200bb). Furthermore, since there is still the unknown factor like in the the old 5c-draw scenarios -- you have a 4 of kind him a straight flush etc. you can bet everything and still lose regardless to how skilled you are.

Additionally, luck still plays a huge role in the short term no matter how skillful you are. You can't bluff/draw every hand and win over the long term. I can't lose to you for thousands of hands and then suddenly flip a switch and make it up in one hand. This is where/why the pool/golf/prop etc hustles work.

Poker is really about the long term unless you go on a short streak (such as perhaps Isil did) and get out. Otherwise the skill factor will eventually kick in; Even if it takes 1million+ hands before you can even start talking about the "long term" there is no real way to hustle anyone over that amount of hands. Poker wisdom is: "The long term is REALLY REALLY long."

I was disappointed there wasn't a punchline about Isildur1 being a bot. Heads up poker is much easier for a computer to play than bigger games. Have the top poker players played against the top bots recently?

Yes there is an annual tournament with the University of Alberta group(http://poker.cs.ualberta.ca/) who is recognized as pretty much the leaders in the field. That said they only play limit I believe since their NL bots aren't yet near good enough to be profitable against the top pros.

Yes, the alberta HU bot plays limit. No limit is a vastly more complex game and the state-of-the-art in HU poker bots couldn't beat a competent low-stakes player.

I would think a million dollars worth of research would be enough to duplicate what the University of Alberta group has achieved. Then you could hire some low-wage people to sit and play limit-holdem online with your unbeatable bot.

I'd be kind of surprised if an elite poker player hasn't already done this.

No way. The members page is currently broken, but http://web.archive.org/web/20080805024322/http://poker.cs.ua... says they have four profs, two consultants, an adjunct researcher, 5 graduate students, and 3 programmer analysts. I can only speak to the costs at the University of Michigan, which I attend, but let's assume they're similar at Alberta. Professors' salaries are around $100k and graduate students cost ~~$50k-$60k before candidacy and probably ~$25k-$30k after. That's at least a half-million dollars per year for the professors and graduate students alone; add the programmer analysts and the consultants and you'll be hard-pressed to get much more than a year of work on the project for your million bucks.

I guess I was thinking that because the research has already been done and they have released the general ideas if not anything resembling an implementation, someone else could duplicate the research faster and cheaper.

If not one million, the same order of magnitude. Conservatively $4MM to do it in two years?

Yeah, I think it's possible (although unlikely) for someone to be able to privately build an AI that can compete at the pro level. As you say, they would have access to all the papers published by AI researchers.

There is always a chance that someone has come up with a new approach. For example, the introduction of Monte Carlo search trees http://senseis.xmp.net/?MonteCarlo to computer was hugely successful. An AI on a 9x9 board can now compete at the Dan level. Obviously that's a long way from top-level 19x19 strength but it does show there is always room for a completely new approach.

Because this story was linked-to from HN, I read the article expecting the punch line would be that Isildur1 is software (or maybe a human+software team). I know there's a group at the University of Alberta that's been working on poker-playing software for some time, and no doubt there are other such groups. I don't know enough about online poker to know if there are mechanisms to prevent the use of software. It seems to me it would be hard to tell.

they have captcha's that show up randomly, but if a human is involved there's pretty much no way to detect what software you're using.

The top poker sites try to make it impossible for bots to play. Which is not to say that it is actually impossible.

As I understand, having a program with statistics and the like open is considered perfectly normal, so "computer-aided human" probably describes either party to this match.

That said, I'm not exactly an expert, having never played even a single game of online poker myself.

that'd be my assumption, too. all of poker players use software like poker tracker or such, but my guess would be that this new guy got a hold of some new program.

what's especially telling (IMHO) was the part where he was 30k down, said brb and proceeded to win much more than he lost. it's almost as if he had a quick'n'dirty bugfix to make.

Indeed, there must be a trick somewhere. First, the guy opens 6 tables and doesn't win. Then there is this "brb" moment where he shuts down 5 tables and start to focus on the last one. The pattern is similar with the other opponents, first he loses, then something happens, then he starts to destroy his opponent.

There is a pattern here. Maybe the guy has programmed a tool/helper that can learn from the playstyle of his opponent really quick (multiple tables at first), making them more predictable and modifying the stats accordingly. I don't know, but something is definitly happening somewhere in the process.

My best guess is that most online players are now "augmented," using at least PokerTracker (hands played databases) with HUDs (heads-up displays that overlay the game window and provide some significant statistical info based on mining hands and odds calculations without player input). I believe, based on what I've read at 2+2 and other sites that even small limit games (say 25/50 cent limit hold'em) are mostly played by players with these tools.

I stopped playing at smaller games almost two years ago because I became convinced I was playing more bots than people - in spite of the fact that I was winning and continuing to build bankroll, I wasn't interested in entering the arms race that I saw the software requiring.

The game sites have a strong incentive to not put in too many safeguards against software automation - they take a small portion of each pot on each hand at a table. The majority of players using software to augment their play during a game are multi-tabling, thus more games/hands = more money for the sites.

I had an interesting exchange with the support staff at one online site, where I was arguing that the simplest thing they could do to discourage data mining was to allow players to use another screen name during sessions as an option. They shot me down pretty quickly.

Even though gathering statistically significant info about a specific opponent requires a prohibitively large number of hands (more than 10K?), most people on HN are highly aware of how we can build algorithms that classify data pretty easily with limited info. By simply deciding that most players fall into a relatively small set of classes, you can probably grab enough info to be positive EV against a new opponent in a relatively small number of hands - this is what you do as a player, but I really feel that adding a software tool to do it tarnishes the game. It makes it a little like playing chess against an opponent sitting there with a laptop with a chess playing program - really the program is what you're playing The human opponent is really just a complicated interface.

But I miss the game, it's a great way to stretch your brain differently than hacking.

Using statistical tracking software is pretty standard for any serious player. It can help a lot playing full tables against 6 to 9 players as stats are used to get an idea of how opponents play, what their ranges are etc.

However I think, these programs are less useful for HU play. When you have only a single opponent and you are concentrating fully on their play your own powers of observation will pick up reads more quickly and consistently than the than the stat programs.

Wow - that was really interesting, despite the jargon. After doing some searches for background information I'm even more impressed. Not to mention shocked by the fact that this top player who has just been knocked off his throne is only 23. I feel poor now :)

One possibility is that Isildur analyzed the current standard of top-level heads-up play and found out that the strategy was exploitable by, for example, playing more aggressively in a lot of cases. If that's the case, standard play will adapt.

In no way is it that simple.

Or you could apply Occam's Razor and conclude that, in a game dominated by chance he simply had a lucky run. There is always a lottery winner, it's just that you can't predict who it will be. The early comments on Isildur's play were that there was nothing special about him but that he was aggressive---precisely the kind of player who would best leverage a run of luck.

I've never played poker online, but I've played some with friends. For me, the fun is in playing against people you can see because you can read their faces. Could anyone explain how online poker is fun without this talking/visual element?

Poker is, at it's essence, a big math problem. You can win by by applying game theory more effectively than your opponents.

Playing online lets you engage in a large, dynamic math game.

Thank you for the insight.

I agree in theory, but not so much in practice. You have people like Chris Ferguson who approach it this way, and it is what the game is at its core, but it's not the way most players approach it.

While I understand the math (I got into playing poker seriously because of an interest in botting -- I found the game more interesting than the programming), I don't approach it from a game theory perspective whatsoever. I run my outs and the pot odds in my head, think about how a win or a loss on a hand will impact me, but I have to weigh that against my instincts and make a decision. My instincts will typically win out.

Even online, you can get very successful reads on people, and you can control the action by learning how the players respond. Because of the nature of online poker -- many, many more hands than in real life, and many more players -- I tend to believe you actually get more inside your opponents' heads (compared to playing in real life), not less, as most people tend to assume.

To respond to the GP, I find online poker fun because it's challenging. There's a constant stream of new opponents, each one with different strategies. For me, it's fun in the same way reverse-engineering is; I have a limited set of information and need to guess the next step before it's taken. The quicker I can get inside the player's head, the quicker I can turn their strategies around on them and either suck them out on my big hands or push them off a small pot with a well-placed bluff.

Fair enough, maybe I shouldn't have used the words "game theory" because, as you point out, the challenge includes game theory but is broader than that.

I see poker as a giant optimization puzzle with a bunch of moving parts. I use a number of math tools--and a few psychological tools--to beat that puzzle.

You can read a lot into playing style online too. How long does someone think before hitting raise? What kinds of hands have they pushed all in with in the past? What kinds of bets/raises do they respond to? When do they fold? Do they call a lot or raise? Re-raise? How do they play in position?

It's fun because you can still read people's minds. You can predict what they have. You know what the outs are, you get people to fold. You can manipulate their minds and effect behaviors.

New development: Isildur loses most of the money he won on the previous four weeks.

http://www.highstakesdb.com/1343-isildur1-loses-over-3m-to-a...

Are there any means of knowing if your opponent is cheating in online poker game?

How can you be sure if he is not supported by bot or if he hasn't some means of influencing what cards are dealt or knowing what cards have been dealt to you?

Is there some poker specific cryptographic tool for that or you must just assume that system you are using to play is completely secure?

Unlike games like chess or backgammon there's currently no AI / bots that can beat no limit holdem or omaha. The University of Alberta and others are studying the game but it will be some years yet before we have a poker equivalent of 'Deep Blue' beating the best players in the game.

The major poker sites are well know and trusted and have proved themselves competent in preventing 'cheating' like what you describe. They're very good at all things security, and you can even get a RSA keyfob to log into your account much like some online banks.

So basically you just trust that poker sites do not have backdoors? That's not something I'd bet my money on.

Everyone who plays poker online shares these same concerns, so it's not surprising that the largest sites care deeply about their reputations (especially since they make so much money playing it straight up).

That said, there has been at least one major incident where one of the original employees of a poker site used a "superuser" account to see his opponents' cards at the high stakes tables and win lots of money. The players figured it out based on his statistically improbable win rate and an accidental release of the full hand history of one of his tournaments.

Congratulations, you win. In fact this has already happened as was the subject of a 60 Minutes piece over a year ago:

http://news.cnet.com/60-minutes-report-how-online-gamblers-u...

I particularly enjoyed this quote:

  "This Grey Cat person was new. And at first, he seemed like a live one. He seemed terrible," Witteles remembers. "He seemed to play crazy. It seemed like he was giving his money away. Except the only thing was, he wasn't losing. He was playing in a style that was sure to lose, but he was killing the game day after day."

Sound familiar anyone?

It seems that discovery of this cheat was mostly due to the newbness of the cheater.

If skilled player possessed backdoor access to poker system and used it wisely, peeking in only from time to time when it's necessary and might go unnoticed he'd be rather undetectable.

> Unlike games like chess or backgammon there's currently no AI / bots that can beat no limit holdem or omaha. The University of Alberta and others are studying the game but it will be some years yet before we have a poker equivalent of 'Deep Blue' beating the best players in the game.

If I were successfully researching poker bot I would not go public with this.

Bot support is nothing compared to collusion.

In an online game, how can anyone be sure that information sharing is not taking place?

Pattern recognition. When you haves 10s to 100s of billions of hands on record it becomes easier to spot anomalies in player strategy that would imply collusion.

Eh?

Collusion over a very short period of time from specific (and easily changed) IPs can make a lot of shared money - it's not clear how (and even if) online poker sites would monitor this.

Collusion isn't that hard of a problem to tackle when people are colluding at a poker table as they act very differently to when they aren't. Since the poker site can see all the hands this becomes visible very easily.

They obviously won't catch all people colluding but they will catch most of them. And just changing IPs won't help much when they ban your account/payment details and you have to arrange another fake identity that checks out at a payment processor regularly.

Listen to Phil Helmuth talk about it: http://www.youtube.com/watch?v=mpMELRn8_l0

Basically, the online casinos are making so much money, they want it to be on the up and up so they can keep taking the rakes.

"He has cemented his name in the annals of online poker as one of the strongest players of all time."

Really? After just one week of running good?

"Since the time I began writing this article, Isildur has lost 2.5M to Patrik Antonius in PLO, cutting his winnings on FTP in half."

What a surprise. Cemented, indeed.

Up to last week Tom Dwan (durrrr) offered anyone a million dollars if they could beat him over 50,000 hands of poker multi-table heads-up. The best in the world lined up to play but durrrr seemed unbeatable. And in one week Isildur1 destroyed him to the tune of $6 million. Even if Isildur goes broke next week he'll always be the one to break the icon that was durrrr.

This isn't very accurate. Antonius accepted durrrr's challenge in early 2009 and the 50K hands are just a little over halfway complete.

http://www.fulltiltpoker.com/durrrr-vs-antonius

The others lining up are still waiting.

I don't follow online hs poker that much, but I think durrrr is far from the unbeatable icon the article portrays. He gets plenty of action online outside of his challenge which suggests many do consider him beatable. Looking at the highstakespokerdb.com stats, it appears he finished up around 6 million in 2008 and is now down nearly 6 million in 2009 (not all to Isildur, presumably).

In turn, Isildur is down nearly 3 million in the past 24 hours alone (after this article was published, which listed his total winnings as "well over 5M"). With variance like that it is way too early to tell how lasting the effects of his successes this week will be.

The $6m is almost all to Isildur1.

Does this mean before Isildur1 arrived he was just about breaking even for 2009?

In any case, my primary objection is that defeating durrr over two weeks does not mean "He has cemented his name in the annals of online poker as one of the strongest players of all time."

Absolutely fascinating.

Tell me - when they casually mention being down 250K or up 1.5M .. is that actual money? Because if so, shit, I need to start playing poker ..

Yes, it is. However, this article only talks about some of the top players in the game. The only millionaires from poker are the top tier, whereas 99% (actually, significantly more) of players are nowhere near that level.

Poker's a whole hell of a lot of fun, and it is possible to make money fairly easily if you play smart, but making large sums of money takes time, patience, and a whole lot of skill.

Agreed, IMHO if you break even you're doing a hell of a lot better than 90% of poker players out there.

Most poker players are doing it because they like the game, and they don't intend to get any money back out of it. Just like the majority of people walking onto the golf course on a weekend has no intention of getting back their greens fee or the cost of their clubs.

As a poker player, golfer, or anyone with any hobby, the ability to regain your investment is normally all it takes. The occasional win taking you above-even is usually all it takes to put you in a great mood. I'm sure having $1M sat on the table, the game is no longer fun as it sounds like extremely hard work.

I play a bit of poker casually, for fun and profit (PokerStars). Mostly $6 SNGs. What software do I need to get? I had no idea add-ons & tracking were so prevalent ..

Get something like tourny manager to manage your money and to watch overall how your playing and stats.

Get something like SNGEGT, SNGWIZ, or SNGPT to run through hands after a tournament to help you determine if your plays were really +EV. This will help you make better decisions in the future when at the tables.

For SNGs you should get SNGwizard to review your push/shove decisions at the end of each session and Holdem Manager to track your results, record your hands, and track your opponents statistics.

Reply on news.ycombinator.com