[sscohen]

In Day Longs, any given hand is played only a few times relative to the total number of entrants.
This prevents people from "peek cheating" in theory. Unfortunately I suspect there is a greater variance in results from player1's board set having more or fewer flat boards than player2's board set.

An extreme example of this is the result I had in 5/31/24's Free Day Long. I had a top on all 8 boards for 65%. By comparison. the overall winner won with 83%. In other words, his/her 8 boards were more swingy than mine.

Don't know whether you can follow this link, but these are my results if you are curious:
https://webutil.brid...sername=sscohen

One simple improvement in my opinion is to throw out tops that are less than a person's score. So for my 65% result, throw out tops that were < 65% and compute my result from the remaining boards.

---

I personally doubt that "peek cheating" (if uncontrolled) would even be the most frequent kind of cheating in individual events. I suspect players gaming GIB's bidding flaws is a bigger issue. (The most recent one that got my goat was someone getting a 100% by overcalling 4n with garbage after a 1-suit opener).

Perhaps BBO could police this with its AI robot?

[pescetom]

sscohen, on 2024-June-01, 11:01, said:

Perhaps BBO could police this with its AI robot?

If BBO had an AI robot with anywhere near policing capacities they would surely use it as a replacement ?

[smerriman]

Yes, the current system of scoring doesn't really work well at all. The official BBO response is that on average the number of swingy hands tends to even out in events like the NABC, but this has no hope of being true in standard 8 board daylongs.

I proposed an alternative scoring system a number of years ago which I still think would be better than the current system (see here, including my second followup comment about 6 below that with the final detail).

As for gaming the robots, that isn't something that should be policed. While of course fixing robot flaws like the 4N overcall bug would help those unaware of that issue, playing well with the robots requires knowing some of its habits, and deciding what counts as 'gaming' and what doesn't is simply impossible - in a sense almost every hand requires some gaming.

[pilowsky]

OP is suggesting something that is done in final high school examinations where results are adjusted to account for the "quality of the candidature" in each subject.
Sometimes termed 'scaling' this addition or subtraction applied to each group ensures that no-one is disadvantaged/advantaged by studying a particular subject.

I'm guessing that this would be difficult to apply to Daylong tournaments because of the small number of boards played by each player.

Also, the complexities involved would probably cause BBO to crash because the person pedalling the bike to supply power to the server would get exhausted, and the string holding it together might break.

Bard said:

Scaling in final exams is used to create a fairer playing field for students taking the same subject but under different circumstances. Here's the basic idea:

Standardization: The goal is to ensure everyone is graded on the same scale, regardless of factors like the difficulty of a particular exam or a teacher's tendency to mark harshly or leniently.

Data considered: Scaling considers various data points, including:

Your raw exam score
Your school's performance on the exam compared to others
Historical performance data for that specific subject
Up or down: Based on this data, your exam mark might be adjusted up (if the exam was harder than usual) or down (if it was easier).

Here's an analogy: Imagine a race where some runners start on a hill (disadvantage), and some on flat ground (advantage). Scaling is like evening out the starting positions to make the race fairer.

The exact process can vary depending on your location. For instance, in Australia, the ATAR system uses scaling for final high school exams [ATAR scaling NSW].

[barmar]

smerriman, on 2024-June-01, 14:45, said:

Yes, the current system of scoring doesn't really work well at all. The official BBO response is that on average the number of swingy hands tends to even out in events like the NABC, but this has no hope of being true in standard 8 board daylongs.

But it's still the case that it evens out in the long run. Some days you'll get the boring hands, other days you'll get the swingy ones. But most of the times you'll get a mix, and the results will be a reasonable representation of how well you played.

And we have lots of daylongs these days. So if you happen to get a bad set of boards on one of them, play another one and it may be more to your liking.

8- and 12-board tournaments will always be higher variance compared to 24 boards and longer. I agree that this design may magnify this a little, but not so much that it's a serious problem. IMHO, it's no worse than having sitouts in a club game when there's an odd number of pairs, and you would have done well on the boards you missed; but it's just as likely that you would have done poorly, and you dodged a bullet. In the long run, they cancel out.