The UNSTOPPABLE Decision-Support Revolution

[Dan Mezick]

The revolution in online Poker Decision Support will not be televised.

Instead, it is coming to a computer near you. It is reducing your online EV day by day, little by little.

The revolution in online poker is a stealth operation. The poker rooms understand this, as do most players with imagination, motivation and hard-core knowledge of computer programming.

Here is an brief and up-to-date example.

PokerTracker, Gametime+ and PlayerView rely upon the player screen name to work. Party allows name changes, and other rooms will follow.

Does this thwart the rapid development of Poker Decision Support Tools?

No.

I've already solved this screen-name problem, with software. It is a very simple process.

1. PlayerA switches names to PlayerB on 2/2/2005
2. PlayerA no longer plays, PlayerB starts playing on or after the same date
3. PlayerB frequents the same games as PlayerA, at the same times of day
4. PlayerB displays similiar stats to PlayerA, especially over time.

There is no where to hide...using the above as input, my programs connect the player-name dots... very quickly. If PlayerA switches to PlayerB and also changes games and time of day, my programs need more time to connect the dots.

Eventually the predictable creatures of habit return to the same watering holes, making the name-parsing and data-based ID process very simple.

There really is no escape once I have clocked at least 20 hours of your play, unless you make a radical change for a very long period of time.

Am I the only programmer on earth doing this? Doubtful indeed.

See my essay on these concepts. Keep in mind I wrote this back in June of 04 and only after using my homegrown HHH tools for many months. Back then, during the Dark Ages, I successfully collected over 500K hand histories by watching the old Party 30-60 Limit Hold'Em 24X7.

(The focus of early efforts will always be on the highest-stakes games, so middle and lower limits are safe for now.)

Alot has changed since early 2004. Hand History Harvesting has become very simple. Party for example, made it ridiculously simple for programs like PT and others to datamine 24X7.

There really is no going back. Roy Cooke in one of his articles about online poker discusses how he thought the 'notes' feature would favor the good players at the expense of the bad and thus potentially reduce online poker room revenue. This is not the case and presently almost every poker room has a notes feature. This is a strong indicator of what is coming in the future in terms of poker room policies for digital Decision Support Tools.

....in a future essay, I will reveal certain emails I received from Stars and Party regarding the use of homegrown Decision Support Tools. I received these emails while developing my online poker programming projects. By definition such programs may be blocked by the poker room at will. Accordingly, I contacted them for clarification of their policies and future intentions regarding the allowance to use Poker Decision Support Tools generally.

The answers are very revealing.

My intention in posting this information is to actively encourage and accelerate poker's inevitable online revolution.

Hand History Harvesting (http://www.pokercoaching.com/poker_coach_resources_article004.asp)

[ QUOTE ]
In the future that is coming, online cards rooms will be forced to supply sophisticated real-time decision support tools to all players for free. Thereafter, players will be playing a all-new kind of poker where every key decision is augmented by software. These developments will be a huge and important milestone in poker's rapid evolution.

[/ QUOTE ]

[Lost Wages]

[ QUOTE ]
I've already solved this screen-name problem, with software. It is a very simple process.

1. PlayerA switches names to PlayerB on 2/2/2005
2. PlayerA no longer plays, PlayerB starts playing on or after the same date
3. PlayerB frequents the same games as PlayerA, at the same times of day
4. PlayerB displays similiar stats to PlayerA, especially over time.

There is no where to hide...using the above as input, my programs connect the player-name dots... very quickly.

There really is no escape once I have clocked at least 20 hours of your play, unless you make a radical change for a very long period of time.


[/ QUOTE ]

<font color="red">I hereby formally challenge you to prove this. </font>

The Ground Rules

1) I play Party 5/10 6-max. Beginning Friday, February 4th, I will play on Empire for a total of 20 hours during my usual playing times. When that is complete, I will switch to playing at Party (where my screen name is Lost_Wages) for the next 20 hours, again during my usual times.

2) When step 1 is complete, I will post a notice to that effect in this thread.

3) You will then have 24 hours to publicly post my Empire screen name.

4) The loser will be required to post a new thread in this forum stating that in fact they did lose the challenge. In addition, the loser will be required to use an avatar of the winners chosing for a period of one month.

What say you?

Lost Wages

[benfranklin]

I have 5 Party skin accounts, and therefore 5 names I could be at Party/Empire/etc. I'm sure many others here have a similar number. Do you claim to be able to find all 5 of those, plus any changes I might make?

[excession]

The POINTLESSLY OBVIOUS thread...

Weall appreciate that poker-tracker type tools will become better and more widespread in the future - what exactly are you predicting?



bots to 'advise' everyone how to play?
that your particular software will take the poker world by storm?

that Party will start providing advice and tracking bots for free?

Your tracking claims seem suspect - play from table to table may change with the texture of the table, external factors, winning early coinflips to dominate (luck), etc

And a player might shift between up to a dozen sites to hunt bonuses and return at irregular intervals

And by the time you have enough data to be sure that player a is player b - then so what? the stats will be the same i any case - 20 hours of logged data is enough to get good reads on anyone anyway

let's see if you really can track 'lost wages' - if not it seems like a case of all mouth and no trousers to me..

[amirpc]

Haha the hilarious thing is that you're going to get workable stats on someone in like just a couple hundred hands with them. So what this means to a PV/PT user is that if someone uses PP's name change feature within a couple hours playing with that person on their new handle you've already acquired enough data to use stats on this person with pretty decent certainty.

Sure if you keep notes on people you'll lose them but man this seems like you're making a pretty big fuss about an imaginary problem.

[Dan Mezick]

Don,

OK. Under the terms of the challenge, I'm pretty sure you can give me the slip, for 2 theoretical reasons, and a 3rd major reason:

1. During Phase1 on Empire, you define 5/10 6max as the game, but during Phase2 when you switch to Party, you don't. No game is stated. It is possible as the challenge is stated for you to go play some other game on Party. Even the best sniffer program would have trouble finding you.

2. You're a PoohBah, so I've got to assume you'd play several styles in Phase2, and at least break even while doing so. Therefore, even if you played 5/10 6max for 20 hours in Phase2, you will not likely get identified by the sniffer in a "new-style same-game" Phase2 scenario.

(1) and (2) above also relate to my caveat:

[ QUOTE ]
There really is no escape once I have clocked at least 20 hours of your play, unless you make a radical change for a very long period of time.

[/ QUOTE ]

...such as changing games and/or styles within game.

3. Even if you relaxed the ground rules under (1) and (2) above, and played the same exact game the same exact way in Phase2, it's unlikely I could identify you within the time constraints of the challenge.

You made a very excellent raise: my little science project is still very much under construction.

As of today, I don't have the goods to call.

This was a very good raise (the avatar aspect was especially menacing.)



Now, perhaps many members who follow your posts will read this thread. And that will tend spread these ideas more rapidly then otherwise.

Those are the results I was looking for.


A peace offering now appears in order: PM me with a postal address to send the CD, and I'll send you the 500K PT hand history DB from the old Party 30-60 limit Hold'Em game.

They were collected during 1Q2004 and 2Q2004.

.

[housenuts]

at the start of every session you should be able to pick a name for that session

[Dan Mezick]

This is an excellent idea. Such a feature is likely very simple for all poker rooms to implement. If they did this, the ability of enterprising players to use opponent data for decision support would be dead in the water.

There are some issues such as duplicate names etc similiar to setting up a Yahoo mail address. If you have ever done this you know it seems every name is taken.

Any market for decision-support software tools would be dead instantly under this per-session naming scenario.

Even so, the potential money rewards involved will keep software developers working on poker decision support, even in the face of this potential threat.

[excession]

'If they did this, the ability of enterprising players to use opponent data for decision support would be dead in the water.'

I log into Party. Change my name. Fire up PT. Start observing 4 tables. Go get a coffee, come back 40 minutes later - sit down at a couple of tables - result - I have reads on most of the table, they have absolutely no read on me.

This seems like I'm still getting value out of PT to me. /images/graemlins/grin.gif

[joeboe2001]

I guess I do not spend enough time worrying about this issue. I tried to set up a second Party account once but it was rejected--yet you say you have 5. Do I have to do anything more complicated than using different email addresses?

[Nick_Foxx]

yeah this is real unstoppable... what about the millions of players who don't play hold'em

mike

[turnipmonster]

hi dan,
I don't buy it for a second, for a bunch of reasons. it's a nice idea, I just don't believe you've got an algorithm that actually does it, and if you did, you certainly wouldn't have phrased it the way you did (using the term "20 hours of play" instead of "X hands"). 20 hours of play to an 8 tabler is radically different than 20 hours of play to a 1 tabler, and someone sophisticated enough to write the kind of software you're talking about simply would not make this mistake in terminology.

--turnipmonster

[Dan Mezick]

It's just a matter of time before all popular games are supported by PT-like functionality by either PT clones or PT itself.

I'd surprised if there is nothing for 7-stud and Omaha8 by November.

[Dan Mezick]

Turnip,

You're right, see my previous reply to LostWages. I do have some crude software going but it's just past the alpha phase of development. Earlier (in 1Q2004) I wrote a complete hand history harvester which run 24X7 to capture 500K hand histories from Party's 30-60 HE game over a 1-month period. That 30-60 game no longer exists, but I am sure the players are still around, likely playing 15-30 now...

I will venture a guess: I will guess that there are dozens, perhaps hundreds, even thousands of very-low-key decision support software projects going on, right now, all over the world.

The new killer online player is likely an INTJ or INTP personality type with a Computer Science degree. This player type is probably not going to do well in a high-stakes B/M casino context, but he is literally born to play online, using PT or a PT clone augmented with his own home-grown decision support tools.

I'm sure there are dozens of specialized tools in use today that are home-grown and giving individual online players a very substantial edge online. The proof-of-concept work I've done confirms it is simple to do. The PlayerA/PlayerB rename problem and others will be solved. New useful stats will be collected.

Eventually, poker online will likely become a specialized form of quantitative/qualitative analysis.

One of the advantages of rolling your own software is that you can create derivative stats and even collect entirely new information.

Take peek at the link to my article (Hand History Harvesting) at the post at the top of this thread to examine likely implications of the coming revolution in online decision support tools.

[turnipmonster]

I hadn't read that far down when I responded, or else I wouldn't have posted.

honestly I fit your profile almost exactly (CS degree, etc) and I think that current player stats/decision support is very overrated, and a lot of times not all that accurate. true, it would be a killer app if you wrote something that gave your opponents possible hands with percentages (i.e. I c/r flop, I have a draw 20% of the time and a made hand 80%), but those stats are going to take forever to converge into anything useful.

also, the problem with matching different nicknames based on playing time/duration/limits doesn't scale particularly well, and that is a significant concern if online poker continues to grow.

--turnipmonster

[Chicanist]

Speaking as a Computer Science major, I'd be very very very surprised if you could actually get this to work with any reliability over 70k players. How rapidly do you think stats converge? And even more importantly, how much do you think they differ once you get to good players?

I can't imagine how 20 hrs of play is even close to enough to correlate these stats enough unless you're talking about running this on a game that has a really small # of people. Hell, if you're running it against eg 3/6 tables how will you even know if somebody switched their screenname? Just harvest tables for like a week and if they don't show up assume that they must have changed names?

good luck though.

[AngryCola]

That was a weak backdown.

Point for Lost Wages.

[illunious]

Seems like an easy solution when you have every players city. Am I missing something or did I just saved you a [censored]-ton of complicated programing?

I don't think getting 8k instead of 4k hands with a pro matters much. The losing players aren't going to change screen names, and they eventually bust out anyway.

[ QUOTE ]
The new killer online player is likely an INTJ or INTP personality type with a Computer Science degree. This player type is probably not going to do well in a high-stakes B/M casino context, but he is literally born to play online, using PT or a PT clone augmented with his own home-grown decision support tools.

I'm sure there are dozens of specialized tools in use today that are home-grown and giving individual online players a very substantial edge online. The proof-of-concept work I've done confirms it is simple to do. The PlayerA/PlayerB rename problem and others will be solved. New useful stats will be collected.

Eventually, poker online will likely become a specialized form of quantitative/qualitative analysis.

[/ QUOTE ]

I agree completely with this.