Back when bitcoin was a mere curiosity, pre crypto currency excitement, well even before that first bitcoin pizza. I happened to have just found this neat piece of software while searching for things related to MPI, parallelism, and that sort of thing. It was called bitcoin, and I had a 32 node OpenMOSIX cluster all to myself for about 12 hours.<p>I kicked off the bitcoin processes let it all fan out to every node, and went to class. Coming back to check on it before I left University that day. I made a mental note to myself about how many "coins" it had ground out of the clusters CPU time and placed in my "wallet", with all the importance one might place on remembering the exact number a computer got on a geek Bench score. If memory serves, it was something around 25,000 bitcoins.<p>At their peak value, that works out to in excess of 30 million US dollars worth of mistake due to lack of desire to back up the results of my idle experiment. To be fair at the time, they were worth less than nothing, I vaguely recall some mention of them 'possibly in future being used as a medium of value exchange' and that sort of abstract thing, and I was running the software out of curiosity. So I'm not bitter.
I once added a backdoor to gain full Admin rights (without corporate sign-off) to a site I was building for a relatively large corporation. Though this was only intended for QA and Dev, I forgot to get rid of this backdoor when I deployed to production, and it was shortly found. A lot of the data that had been input was mangled and broken (or just outright deleted) and not backed up, which forced a lot of the creative people to redo several days of work.<p>I was fortunate enough to have a boss was pretty understanding, and I was not fired, though I genuinely felt he would have been justified in doing so.
The most expensive mistake I've made is not focusing on a single goal, task, technology, or stack. I keep changing my goals. Because of that I have learnt nothing today. Any idea to get rid off this habit?