"Just use transactions" is not sufficient to solve concurrency problems under load.
I described a real-world case where it was not to demonstrate the point.
You were "Presumed innocent until proven guilty", but the proof was pretty easy to produce.
As for my parting comment about whether you had relevant experience, that is an educated guess at best, which you could easily confirm or deny. However I should point out that most programmers don't have the kind of experience that I described, so it is no insult to suggest that you don't either. (Neither did I until I moved to LA.)
Cheers,
Ben