Reminds me of this: https://en.wikipedia.org/wiki/Fallacies_of_distributed_computing https://en.wikipedia.org/wiki/Fallacies_of_distributed_compu... Part of th

Reminds me of this: https://en.wikipedia.org/wiki/Fallacies_of_distributed_computing https://en.wikipedia.org/wiki/Fallacies_of_distributed_compu... Part of the reason I find distributed and concurrent computing so fascinating is that we lose so many assumptions about how computers are supposed to work; ordering guarantees, data actually arriving to destinations, fairness, etc. While it is of course possible for data to get lost or mangled in a single-threaded, non-distributed application, it's rare enough to where you often don't need to plan much around it. Once you add concurrency, and especially if you make it distributed, you suddenly have to plan around the eventuality of the messages not arriving, or arriving in the wrong order, or arriving multiple times. Personally, I've gotten into the habit of PlusCal and TLA+ for everything, and modeling in the assumption that stuff might never arrive or arrive multiple times, which can work as a very good sanity check (though it's of course not a silver bullet).