I agree with you. A long time ago, those of us "in the know" techies could parse the difference like it was a native language. When talking anything but computers, it was always the SI of 1000. When talking about computers, it was always 1024.

I think the masses were confused and the SI purists felt their SI prefixes were being corrupted. So they made a distinction/standard between binary numbering system prefixes and decimal numbering system prefixes.

I hate it. Feels wrong because I'm old and set in my ways. People like me are confused because we still use the old nomenclature, and when someone else uses the old nomenclature (when talking about computers), it's ambiguous to us because we don't know which numbering system they are using (e.g., binary as opposed to decimal). I still have to ask and half say binary and half say decimal.

I suppose if they're teaching it in high school and college it'll become native soon enough, if it hasn't already with the next generations.

Relevant xkcd

I don't understand why we can't just make 1 G=1 G, ditch the silly contex-sensitive-prefixes nonsense, and call it a day.

I blame memory manufacturers. They're the ones who started this whole 1 GB sometimes= 1024 MB bullshit.

An important thing to note about this is that as we go up exponentially the error between GiBs and GBs increases. A kiB is only 2% more than a kB, but a TiB is 10% more than a TB. So using them interchangeably is increasingly misleading.

Also, there are many cases in computers where it doesn't really make sense to fuss about binary. Like, an HDD is a spinning piece of metal, the number of bits it can store has no binary constraint.

Fun fact: the old 3.5" floppies that were marketed as 1.44 MB were neither 1.44MiBs nor 1.44 MBs, but some weird hybrid mash-up unit.

To expand on this: people sometimes use SI prefixes to refer to 1024 units, but it's just wrong. A kilometer is 1000 meters, a kilogram is 1000 grams, and so on. If we were to re-define these prefixes for specific disciplines things get much more complicated very quickly.

@skullgiver @LaughingFox It's not programmer laziness at all, RAM modules' size has to be of a power of 2 on most platforms because of various assumptions the CPU makes in memory alignment and memory bulk reads for performance reasons. Processors don't interact directly with the flash dies, so it's fine for them to be of the size they feel like provided the controller knows what it's doing.