Hacker News new | past | comments | ask | show | jobs | submit login

What Engineering tech/AI tech do you think could make the process more thorough but not proportionally expensive?



In crypto the auditing process is somewhat more sophisticated. They scan the contract for similarity to known scams and analyze it for possible backdoors. They also do due diligence on the promoter of the contract ("fully doxxed").

In reality of course all this work could have been replaced by def is_fraud():return True

And the accuracy would probably increase. Crypto fraud has the beautiful property that the people being defrauded actively defend the fraudsters. Moreover, in a lot of cases it isn't technically fraud since the contract is upfront about what it does but at the same time it is very exploitative but that doesn't matter to crypto people


how do you technologize intent-detection? maybe chatgpt-x could do it, but that's the crux.

i am NOT haying pattern recognition won't help, search for audit software and you will see each of the big four has specialized software. (here is EY's: https://www.ey.com/en_gl/audit/technology)

the problem is the issue of perverse incentives, IMHO. Audit takes a butt load of time and money, and disrupt business while they do their thing, and pays peanuts frankly... and audit firms earn more from associated services, contracts which they can earn if they don't bother the management too much.

yes, there are a dozen caveats and stuff, but frankly, the issue comes down not to technology but to people. The same network of people are in the few audit firms, and the spin out to join companies sometime later, who hire the same few audit firms, and so on.




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: