Hacker News Reader

Comment by rossdavidh | Hacker News Reader

So, this is all interesting for chess, but I wonder if we should be discussing more the broader implications. How much does this happen during high-stakes tests (professional qualifications, etc.)? How much does this happen during job interviews? How much does this happen during political debates? How much does this happen during any high-stakes activity where the person is supposed to be thinking through it on their own?

View on HN

What I find interesting about this case are the implications around using what I would call "cognitive augmentation" devices. I can't help to compare this situation with the revolution that's happening with art and ai.

I suspect the organizers of Chess.com are at the beginning of a cat and mouse game that will produce more sophisticated detection algorithms, which in turn will help develop more sophisticated chess engines.

I bet that for lovers of chess, who want to continue to enjoy their passion, this is a critical problem to figure out.

However, can we expect that in games, sports, arts, humans will continue to compete with each other without any forms of artificial aid? Or should we instead explore now what augmenting human's cognition means for our culture? Starting now might be the only way to creating a framework to ensure success (however we want to define it) in the long run.

Well, people on blind have admitted to cheating during algo interviews. Not sure how prevalent it is, but a lot of people openly admitted to it. Guess it shouldn’t be surprising given the nature of these interviews and the money at stake.

I also know from my time at school, cheating is fairly prevalent there - even at top rated universities. Lots of universities have developed fairly sophisticated statistical tools for catching cheaters, similar to what chess.com has.

The literal majority of the class cheated at my college. It was unusual not to cheat.

I'd be willing to bet that the prevalence of cheating is roughly equal regardless of the discipline. Let's say 1% of participants are cheating at any given time (I have no idea if this is a high or low estimate).

For better or worse, this could represent a metric for gauging how well that discipline is at detecting active cheaters.

I'm quite skeptical that it's equal as that would mean that cheating is not affected by, nor affects community/culture. I think we would find, if we explored this deeper, that occurrences of cheating are clustered as they feed off themselves in a developing culture of cheating.

I recall the bagel measure of cheating found that different industries had significantly different rates of cheating. This was a business where office break rooms had self-serve bagels, and you paid for it by putting money into a money box, on the honor system. Most people were honest, and so the business just dealt with the 10% or so who took a bagel without paying, as a cost of doing business. After a while, he realized that he had a measure of the general level of honesty by industry. I think Freakonomics wrote about this.

I think clusters of cheating would be especially prevalent in cases where there was a finite pool of positions available, and more people than that who wanted to get a position. If you know others are cheating, it's a pretty strong rationalization for cheating yourself, and the more prevalent the cheating, the stronger the rationalization.

Reply on news.ycombinator.com