Ask HN: What is the thing you've built that you regret the most?

Critically (compared to cash) it allows remote transactions. Cash requires presence which limits the customer base geographically, and carries enormous personal risk. (you're already transacting with a criminal, do you want to be alone with him as well?)

The killer app for crypto is illegal activities. (which may in some cases be moral, but nevertheless are illegal.)

As a side note, the volatility in the crypto value, caused mostly my legal speculators being exploited by scammers, hurts the utility of crypto in all contexts, including illegal ones.

As we scrambled to create something of value and keep the lights on, I (unintentionally) built and highly optimized a free-trial funnel for a Saas service according to a "gym-membership" model; ie, our entire revenue stream depended upon tricking people into submitting their credit cards and charging them for months when they forget to cancel (or couldn't due to the complicated cancellation funnel). Once someone hit gym-membership status, we would pause all emails, reminders, etc (on CEOs design) so they would forget about us and let their card be charged for years. People at our company would fight against these tactics, but leadership's only focus was AB testing the hell out of the funnel to continually increase subscriptions and impede cancellations.

To combat the inevitable high charge back rates we eventually encountered, our staff would purchase pre-paid gift cards at corner markets and we programmatically submitted multitudes of tiny transactions through out the day to skew the chargeback rate to an acceptable place; this was the CEOs idea again, rejecting our ideas of selling things people actually wanted.

It was a house of cards, but the success of monetization was leveraged to land further contracts with governments (that we could never fulfill) until it all came crashing down. I left long before then on principle.

The entire venture was revealed to be a complete mess from day 0. From the start, this outfit threw its entire batch seed into google ads to drive "users" and feign growth to pump up the valuation on demo day, landing a couple million in investment for something that had no real value. Hm. It seems that a system was crafted here to pick winners and losers, and the company responded by gaming it in every way they could.

I remember a conversation ages ago about how you couldn't really get a (common?) lisp running properly, irrc due to limitations in the way CLR modeled classes amongst other things, but FFI came up there too.