Poker:the rake, the variantsand the bots problem

Poker is the one casino format that is also a game of skill against other people, and that single difference shapes everything about how it has been built into software.

What the house is actually selling

In every other format on this site the house is the counterparty: it wins when the player loses, and the maths guarantees the margin. In poker it is not playing at all. It takes a rake (a cut of the pot or a fee per hand) and is indifferent to who wins.

That changes the product. A poker room does not need a house edge, it needs traffic: enough players at enough tables that a game is always running. Where a slot release competes on theme, a poker room competes on liquidity, and the record of the online poker business is largely a record of rooms merging to pool it.

The variants and what separates them

Texas hold'em dominates because it is the easiest to watch. Two hidden cards per player and five shared ones means a spectator can follow a hand without seeing anyone's cards, which is not true of the older stud games where most of the information is private.

Omaha changes one number (four hidden cards instead of two), and the consequence is disproportionate. More starting combinations means stronger hands are common, pots are contested further, and the variance climbs. Draw and stud variants persist mostly in mixed-game rotations and not as standalone rooms.

Poker built into video games

The interesting cases are the ones where a hand is a real game, not a minigame reached through a menu. A poker table sitting inside a world, with the same characters and the same economy as everything around it, behaves differently from a standalone poker app: losing costs something the rest of the game recognises.

Most implementations do not go that far, and the reason is not laziness. A hand of poker takes minutes, involves opponents who need to bluff convincingly, and cannot be balanced the way a combat encounter can. Writing an opponent that plays badly in an interesting way is a harder problem than writing one that plays well.

Why the bots problem is structural

Because poker is skill against other people, software that plays it well takes money from other players and not from the house. That makes it a policy problem for the operator rather than a maths problem, and it has no clean solution: detection is adversarial, and every measure that catches automation also inconveniences the humans.

No other format on this site has this shape. A bot playing a slot machine is just a player losing at the advertised rate.

Why the online game is not the live game

Two differences do most of the work, and neither is about the cards. The first is speed: an online table deals several times faster than a live one and a player can sit at several tables at once, so the number of hands seen in an hour is an order of magnitude higher. Variance that would take a live player months to work through arrives in an evening.

The second is information. Live poker leaks through posture, timing and hesitation; online poker leaks through bet sizing and timing alone. That narrows the channel and makes it measurable, which is why the online game developed a statistical literature and the live game mostly did not.

Both differences push in the same direction: the online format rewards volume and record-keeping over reading a table. That is a different skill wearing the same name.

Which is worth keeping in mind when reading anything written about poker strategy: almost all of it now describes the online game, because that is where the hands are and where the data accumulated. The live literature is older and much thinner.

What this section leaves out

No room comparisons, no rake schedules, no traffic figures and no strategy instruction. The format and its history are the subject. The wider format record sits at casino and table games.