Yesterday, the Commerce Casino in Los Angeles began experimenting with a new format, called “Time’s Up.” As the name suggests, the concept is that the tournament runs for a fixed length of time, rather than until only one player remains. The time limit and structure are chosen such that the bubble has burst by the time the tournament finishes. Those who’ve busted before the time limit but after the bubble will receive their prizes according to the normal payout structure, but those who still have chips will instead have their stacks counted, and their prizes awarded according to their position in the chip counts.

Correction: It turns out that the payouts were based on an ICM chop rather than position in the chip counts as is implied in the video. This negates some of the unfortunate side-effects mentioned later in this article, and makes for a relatively normal tournament, just one in which players are forced to chop at a certain point, probably slightly earlier than they would ordinarily.

There were actually two such tournaments run, as part of the California State Poker Championship, with $350 and $175 buy-ins. Both had similar structures, but the higher buy-in event began with more chips (50,000 versus 25,000) and had a 12-hour time limit, while the $175 event gave the players only eight hours. The events likewise had their start times staggered by four hours, such that both would wrap up at the same time. There’s no word yet on who won or how the tournaments were received, but the $350 event drew 167 players, the exact number required to meet its $50,000 guarantee, so on that level it was neither a huge success nor a total flop.

Trying for the best of both worlds

It’s pretty easy to see what the goal is with this tournament format. Like a lot of experimental structures, it’s an attempt to satisfy players’ mutually contradictory desires for a “good” structure allowing for deep stacked play, and an event which wraps up in a reasonable amount of time, since multi-day events can be extraordinarily exhausting, particularly for amateur players.

Indeed, the blind structures employed by Commerce Casino are of the “fast-slow” variety that’s been gaining traction lately, beginning with extremely deep stacks (1000 big blinds, in the case of the $350 event) and turbo blind levels (15 minutes to start), then slowing down as the tournament progresses, ending up with 40-minute levels, and then a final level with no time limit, but rather the number of hands fixed at 40. This structure has the same goal in mind, providing lots of play at the beginning and end of the tournament, but progressing rapidly through the pre-bubble stages.

Online sites such as PokerStars have tried similar formats, with mixed success. These tournaments generally pay players out based on their share of the chips in play, however, rather than their position in the chip counts. That makes them quite simple, strategically; since chips have a linear value in that format, they’re effectively just cash games, but with freezeout mechanics, table balancing and escalating blinds.

Weird effects in the late game

In the Time’s Up format, on the other hand, the cash equity versus chip count curve fairly closely resembles that of a regular tournament up until the bubble, but gets very bizarre thereafter. That is, chips are initially fairly close to being of equal value to everyone, and players can play a normal strategy through the early levels. Towards the bubble, the usual ICM considerations come into play, as short stacks attempt to survive until it bursts, while deeper stacks exploit that dynamic to accumulate chips.

As the tournament’s time limit approaches, however, chips begin to lose their intrinsic value, and instead gain value based on the player’s position relative to others. If you imagine a bunching of players with similar chip counts, but with relatively larger separation from the next higher and lower players, it’s clear that those with lower counts have a lot to gain and little to lose as the end of the tournament approaches. Conversely, those on top of the log jam would like to stay there, and are more incentivized to avoid losing chips than to gain them. The most extreme case comes in the final hand, when the player with the fewest chips should never fold unless someone else is likely to go broke, and the first place player should fold almost any hand, unless second place is extremely close in chips.

Problems with the format

These considerations lead to some potential problems with the format. Any time chips are worth more to one player than another, collusion becomes possible. In a normal tournament, that tends to take the form of a deeper stacked player helping a friend survive the bubble or make an important pay jump, but many more possibilities present themselves.

At the tail end of the tournament, pay jumps are achieved through chip accumulation rather than survival, so a pair of players may be able to increase their combined payouts by moving chips to the player with the most to gain by moving up in the standings. For example, most tournaments’ structures will give more to 1st place than to 3rd and 4th combined; in a conventional tournament, the medium stacks want to avoid all-in confrontations with one another, but here, if 3rd and 4th have enough combined chips to overtake the leader, they will want to try to get into a race with one another, particularly if they have a deal in place to share their winnings.

Even if none of the players remaining in the tournament have such a deal, all sorts of undesirable behavior is encouraged – making one’s stack hard to count, getting up from the table frequently to count those of others, and so on. There are also going to be a lot of situations arising in which two players haven’t explicitly made a deal to collude, but end up doing something that looks like collusion because they’ve both realized it’s in their mutual best interest.

Fuzziness is the game designer’s friend

These kinds of fiddly edge cases are something that comes up frequently as a problem for game designers. Very often, a given game mechanic works very well most of the time, but leads to undesirable, un-fun gameplay on the final turn, or in certain situations, particularly between serious, experienced players who will do whatever they can to win.

The way to resolve this problem is often to introduce some kind of fuzziness to the game. If, at time T, there’s a huge difference between having N of something or N+1 of it, and this is what’s causing the problem, then one possible solution is simply to make it hard for players to determine the exact values of N and T. That is, you ensure that there’s some fuzziness or uncertainty involved in knowing beforehand exactly how much of that thing the player needs, or when it’s going to be counted. The value of the quantity in question will still ramp up quickly when T approaches and the player is close to the expected value of N, but the corners of that sharp step will be rounded off.

In a way, some of the weaknesses of live poker turn out to be a strength when it comes to this format. In particular, the lack of exact chip counts is vitally important, which is likely why this specific version of the timed tournament idea hasn’t been tried online. If it’s not clear what I mean, here’s a picture:

The graph on the left is what a pay jump for this tournament looks like when the opponent’s chips are known exactly. On the right is what the same pay jump looks like for the player, when she only has an estimate of the opponent’s stack size, with an expected error of roughly +/- 5%. Her chips still gain a lot of value as she approaches the opponent’s estimated stack, and decline in value again thereafter, but we no longer have the situation where one chip is worth zero to her, the next one is worth a huge amount, and the one after that is worth zero again.

How else can we increase fuzziness?

I suspect that natural uncertainties in chip counts won’t be enough to make the format work in the long term, however. For one thing, the standard tournament payout structure places huge emphasis on the top few pay jumps, and this is the point in the chip counts where players stack sizes tend to be most spread out. If the separation between the players in the top few spots is greater than the uncertainty those players have in eyeballing each others’ stacks, there’s still going to be a lot of weird stuff possible in the final few hands.

One thing that I think needs to be done is to flatten the structure out, so as to increase the importance of pay jumps for those in the middle of the pack and decrease it for the top few players. That would decrease the incentive for collusion even if exact chip counts were available, but it will also concentrate more of the expected value of chips in the part of the payout structure where players’ totals are tightly spaced and uncertainty is greatest.

Another idea would be to introduce uncertainty into the length of the tournament as well as the chip counts. As they’ve run it at Commerce Casino, there are exactly 40 hands played in the final level. One improvement I would suggest is to go hand-for-hand after, say, 38 hands, then start flipping a coin after each hand is completed, to determine whether play ends or another hand is played. No amount of chips will ever be entirely valueless to anyone in that case, because they can never be sure whether or not there will be another hand played in which those chips might become relevant.

Problematic but promising

Despite the potential for bad outcomes and controversy, and the likely need for some tweaks, I’m glad to see this kind of tournament being tried out, and hope such experimentation will continue. Aside from the intended goal of balancing the pacing and duration of play, the format is also likely to be a lot of fun simply because it incentivizes chip accumulation.

Money bubbles, final table bubbles and other high-ICM spots are not much fun in general, especially for short stacks and recreational players, because they force such a tight style of play and a lot of angst over lost chips and bust-outs. A format which encourages players to chase payouts aggressively rather than playing a waiting game has a lot of potential to be fun; this is why, for instance, Progressive Super-Knockout tournaments have proven so popular on PokerStars. Unfortunately, keeping track of bounty sizes makes that format impracticable in a live environment, but some version of a Time’s Up-like tournament might eventually prove to be the way to produce a similar dynamic in a live event.

Alex Weldon (@benefactumgames) is a freelance writer, game designer and semipro poker player from Montreal, Quebec, Canada.