Hacker News Reader

Comment by tempestn | Hacker News Reader

It's pretty easy for good players to recognize other good players. And since the house takes such a large cut, the only way for pro players to have positive expected value online is to seek out games with poor players. So even if they couldn't recognize the bots as such, they would see them as tough players and avoid them.

That said, I suppose it would be possible for the bots to become so prevalent that all this sort of opportunity is effectively used up, so the return vs time and risk for a human player is no longer worthwhile. (That already happened long ago for most players, as the initial online poker boom faded and most casual players left.)

On the other hand, all the major platforms have terms prohibiting using bots, so their numbers might be sufficiently limited to prevent that scenario.

View on HN

It's my understanding the big sites have some pretty sophisticated bot detection systems, so in theory a bot that would be successful at beating online poker couldn't be a huge winner, it'd presumably raise too many red flags. However, if it were a near break-even player, with dozens, if not hundreds, of instances running at any given time, it's going to slowly grind out a substantial figure. You'd also have to take into account that the sites are monitoring things like reaction times to bets and raises, hand range consistency, etc. I'm not a coder, but it seems like it'd be a tremendous undertaking to code a bot that would be a substantial threat to players. Then again, maybe I'm naive about the level of scrutiny the poker sites are employing.

I guess the real threat isn't a "bot" but something in the way of a program that interprets the data on the screen real-time and whose output instructs the player of the "optimal" play, given the circumstances. How the hell would you deter that as a site operator?

Pretty sure I've read a long time back on 2p2 that large consistant winners on certain sites have been asked to submit camera footage of their play with a clear view of screens and inputs. So this is probably something that companies like Pokerstars have been dealing with for years already.

It would be pretty easy to hide something signalling you on what to play from cameras.

True, but ultimately if they're unsure they'll just ban you from the platform anyway. Consistent, winning players aren't really where they make their money, and they're free to ban anyone they like. (I realize technically they take a cut from all players, but more money gets sloshed around for them to skim off of if winning players aren't removing it from the system.)

No, I think your earlier example of a swarm of just-above-break-even bots would be much more difficult to combat. Even if they can be detected, the anti-detection countermeasures can evolve, turning it into an arms race. Anything you can model in your bot detection algorithm, the bot-maker can model too.

Reaction times ought to be one of the easiest things to fake. All it would take is a bunch of monitoring of large numbers of games to create a nice model of real player reaction times, which in all likelihood are normally distributed anyway.

Not normally distributed, as negative reaction times are unlikely. You could use log-normal, but I believe that a mixture of exponential and gamma tends to be used by reaction time researchers (search ExpGamma).

negative reaction times are unlikely

Oh, right. I was thinking along the lines of 100m dash, where people often do have negative reaction times (which we penalize as false starts).

In poker we don't have much of an incentive to react instantly to any play.

That was what I was thinking, the bot augmenting a human's playing ability rather than playing itself.

>You'd also have to take into account that the sites are monitoring things like reaction times to bets and raises, hand range consistency, etc.

You might be surprised by the lengths people go to in order to bypass bot-detection just for ordinary games. All of the things you mentioned are pretty standard. Considering there is serious money on the line here, I am positive that plenty of poker bots will be virtually indistinguishable from professional players, if they aren't already.

The same argument of money being on the line applies to the detection. Poker software is already pretty damn impressive with its tracking. The online casinos actually stand to lose more money than the bot creators could make, so the detection has a greater incentive, and is likely to triumph.

They only lose if there are less plays, surely? I assume they take a cut of all winnings, they're not putting up stakes.

Yes, I'm assuming that if bots work their way into everyday online poker that people will stop using it, so there would be less players.

Poker sites mainly detect bots based on their login times, number of tables, time per action, etc.

A successful bot shouldn't get caught for "playing like a bot" because the moment it's actions are that predictable it would presumably no longer be effective.

But it will get caught for operating like a bot. So, don't run it 24hrs a day. Sites also randomize things to keep bots at bay, even card imagery.

If your performance and success drops whenever they randomize something that gives the bot false inputs, then you might get caught.

Inputting all of the poker events manually would be really tedious I'd imagine.

Of course, if you're winning millions, they can interview you about your poker history and how you got so good.

It sounds like easy money, but probably not.

Just play as you normally would, with the bot advising moves from the laptop next to you.

Right, but the bot needs to know who is in what position what the bets are, who folded, etc. Try inputting all of that information manually to the laptop next to you and you'll quickly get frustrated. Online poker is a fast game with lots of data points.

TensorFlow, PyTorch, Caffe, Keras, MXNet, and OpenCV could copy the game if you split the video input for the player and the bot.

Yes, but see my previous comment.

People have tried it and online poker sites know they've tried it, so they'll randomize images and other data. If you take a dive when the randomizations are triggered and outperform otherwise good luck trying to collect your winnings.

An external camera with Image processor does that

Not to mention, if you get caught, there could be worse consequences than just having your account locked. The site could (and likely would if the scale was significant) sue you for not only all your winnings, but damage to their business. They would likely win (since you're flagrantly breaking their terms of use contract), and bankrupt you.

Edit: In fact, if we're talking worst case, circumventing their anti-bot restrictions would presumably be illegal under the CFAA. So if you're in the US you could even be charged criminally, although I expect in reality that would be less likely.

One of the professors I used to work with some years ago was involved in stylometry research on human-computer interactions such as keystrokes and mouse input (for example, to determine if a user who had authenticated successfully earlier is not the same person currently typing based on keystroke cadence and pattern analysis - e.g. if someone sat down at an unlocked workstation and started typing, you could detect it and force them to reauthenticate).

It would probably be possible to figure out the types of detection being performed by the poker sites and use adversarial training methods to train a machine learning solution to mimic human input patterns. Or, more pragmatically, have the bot analyse the state of the game and give orders for a human to perform at their own natural pace.

> So even if they couldn't recognize the bots as such, they would see them as tough players and avoid them.

The problem would be if i was a pro i would rather run 1000 bots than play myself. Which means the only players left are AI and fish. Once the fishies learn of this fact, they will abandon in drove.

It's all gonna go back to live poker soon.

Reply on news.ycombinator.com