xlfe

Thank you for your subsequent reply, I appreciate it and don't disagree with what you're saying - no one owes me any of their precious time nor attention.

I'm writing publically for a few reasons - one being to help arrange my own thoughts (which this capital without labour piece did), another being so that people (or bots) who do want to understand my thought process - perhaps because they're thinking about working with me - have a chance to see into my thinking on a particular subject. I accept that's going to be a pretty small audience!

I'm not suggesting the future isn't dystopian, where capital is further enriched, in fact it's pretty clear it does look that way. For me though, I'm figuring out what's my place in that. I don't honestly think I could affect the course of history here.

But, just to say, your feedback did resonate with me, and I've already taken it on board, and will endeavour to produce (more of) my own mistakes in the writing I publish.

Cheers

Author here - yes I use AI to help me draft and research, but it's still my writing, and I spent many hours thinking through the arguments and what I am saying.

The "written with AI" charge is used so indiscriminately on HN now that it mostly functions as a way to dismiss an argument without engaging with it.

If there's a substantive point you disagree with, I'd genuinely like to hear it.

The author of this article (Paul) appears to ignore the law of large numbers - while Paul's arguments against using the figures from Backblaze as measures of drive reliability would hold if Backblaze's analysis was based on a small sample size, because their sample size is relatively large, Paul's criticisms don't hold. Paul's argument is that any analysis of hard-drive reliability should control for factors like vibration and temperature as these vary in all environments - but in fact, because Backblaze has taken large (enough) sample of drives, these variations are already controlled for (ie cancelled out). Sure the consumer is probably never going to experience a failure irrespective of whether the buy a seagate or WD drive - but that's not the point - the point is that the long tail of the distribution, WD and Fujitsu beat Seagate for reliability. That is still valuable information for people, even if they aren't going to be using their drives in the same type of environment as Backblaze.

Who knows? All I am doing is working from the data available.

I've tried to be clear about the methodology. My guess is that most users just reuse a small number of passwords. Anecdotal evidence supports this.

If you have a suggestion for how to derive a better _estimate_ from available data I'd be very interested to hear it.