I already had other (newer) computers being backed up, and this was a courtesy email to tell me I had failed to stop paying for monthly backups for a computer I

I already had other (newer) computers being backed up, and this was a courtesy email to tell me I had failed to stop paying for monthly backups for a computer I had not opened (and thus synced) for years. He emailed me because the version running on that computer was considered so unstable as to be unreliable. Yes, they could have proactively canceled my bill. Maybe they would have! We'll never know, because I responded to this email right away. He responded to my questions in a lengthy email, which I'll paste again here because I had the foresight to ask in advance for future permission to post publicly. === Hi Robbie, thanks for getting back to me! It's hard getting through the email marketing overload. :-) > Would you be amenable to holding off on deleting the backup > for two weeks (instead of one) No worries, the most important part is that you are aware the computer is no longer actually performing fresh backups of new data. We can wait a while. > would you mind if I published your email publicly? Go ahead, and one of our marketing guys (Yev) already saw your tweet go by, it made everybody in our office laugh too. I'm off in my corner watching logs and emailing the random old customer, they didn't know I do this. :-) One of the things I've been amazed at doing backup software is how many assumptions I made at the start that aren't true and the MODERN Backblaze software works around: 1) assumption: your disk on your laptop will reliably store bytes or fail. Wrong, it can throw random bits, undetected unless Backblaze keeps its own parallel check sums. 2) assumption: laptop RAM works reliably. Wrong, I think there is a reason servers run ECC RAM, and I'm confused why many laptops choose not to. I guess saving $5 is more important than data Integrity? 3) assumption: the HTTP posts either work or fail. Wrong: WAY MORE OFTEN than the file system are undetected silent data corruption. Solution: client end to end SHA hashes. 4) assumption: customers are sane and won't "edit" the Backblaze internal data structures sitting on the customer's laptop. Nope! If customers are computer beginners they tend to behave nicely, and the most advanced customers are very careful, but there is this crazy middle ground who know enough to be dangerous to themselves. :-) I really really think that the most modern version of Backblaze (2.3.1 and later) is quite stable and might run flawlessly for a decade on a computer, but some of those early versions like anything starting with a 1.0 or 1.5 scare me because there are too many assumptions in them, not paranoid enough. Yes, the old versions WILL WORK for 99 percent of customers, but at our customer numbers that means every day we have to explain why a restore failed to some poor guy who chose to edit his Backblaze data structures corrupting his own backup and we didn't detect it soon enough. Our most recent client versions are deeply, profoundly paranoid software that basically assumes your laptop is right smack dab in the middle of melting down, throwing disk and RAM errors as an insane laptop owner not only ignores the warning signs of impending data Armageddon, but fights with Backblaze trying to defeat it from backing up because the customer is worried Backblaze is "wasting his bandwidth". I feel good about this, it's like my whole career was training up to write this incredibly important piece of software. -- Brian Wilson. CTO Backblaze brianw@backblaze.com