Comment by parasense
10 months ago
> This is par for the course I guess, and what exhausts folks like marcan. I wouldn't want to work with someone like Ted Tso'o, who clearly has a penchant for flame wars and isn't interested in being truthful.
I am acquainted with Ted via the open source community, we have each other on multiple social media networks, and I think he's a really great person. That said, I also recognize when he gets into flame wars with other people in the open source social circles, and sometimes those other people are also friends or acquaintances.
I can think of many times Ted was overly hyperbolic, but he was ultimately correct. Here is the part of the Linux project I don't like sometimes, which was recently described well in this recent thread. Being correct, or at least being subjectively correct by having extremely persuasive arguments, yet being toxic... is still toxic and unacceptable. There are a bazillion geniuses out there, and being smart is not good enough anymore in the open source world, one has to overcome those toxic "on the spectrum" tendencies or whatever, and be polite while making reasonable points. This policy extends to conduct as well as words written in email/chat threads. Ted is one of those, along side Linus himself, who has in the past indulged into a bit of shady conduct or remarks, but their arguments are usually compelling.
I personally think of these threads in a way related to calculus of infinitesimals, using the "Standard Parts" function to zero away hyperbolic remarks the same way the math function zeros away infinitesimals from real numbers, sorta leaving the real remarks. This is a problem, because it's people like me, arguably the reasonable people, who through our silence enable these kind of behaviours.
I personally think Ted is more right than wrong, most of the time. We do disagree sometimes though, for example Ted hates the new MiB/KiB system of base-2 units, and for whatever reasons like the previous more ambiguous system of confusingly mixed base-10/base-2 units of MB/Mb/mb/KB/Kb/kb... and I totally got his arguments that a new standard makes something confusing already even more confusing, or something like that. Meh...
> Ted hates the new MiB/KiB system of base-2 units, and for whatever reasons like the previous more ambiguous system of confusingly mixed base-10/base-2 units of MB/Mb/mb/KB/Kb/kb
Here's my best argument for the binary prefixes: Say you have a cryptographic cipher algorithm that processes 1 byte per clock cycle. Your CPU is 4 GHz. At what rate can your algorithm process data? It's 4 GB/s, not 4 GiB/s.
This stuff happens in telecom all the time. You have DSL and coaxial network connections quantified in bits per second per hertz. If you have megahertz of bandwidth at your disposal, then you have megabits per second of data transfer - not mebibits per second.
Another one: You buy a 16 GB (real GB) flash drive. You have 16 GiB of RAM. Oops, you can't dump your RAM to flash to hibernate, because 16 GiB > 16 GB so it won't fit.
Clarity is important. The lack of clarity is how hundreds of years ago, every town had their own definition of a pound and a yard, and trade was filled with deception. Or even look at today with the multiple definitions of a ton, and also a US gallon versus a UK gallon. I stand by the fact that overloading kilo- to mean 1024 is the original sin.
> Another one: You buy a 16 GB (real GB) flash drive. You have 16 GiB of RAM. Oops, you can't dump your RAM to flash to hibernate, because 16 GiB > 16 GB so it won't fit.
Right but the problem here is that RAM is produced in different units than storage. It seems strictly worse if your 16GB of RAM doesn't fit in your 16GB of storage because you didn't study the historical marketing practices of these two industries, than if your 16 GiB of RAM doesn't fit in your 16 GB of storage because at least in the second case you have something to tip you off to the fact that they're not using the same units .
I want to say that I am thankful in this world that I am a truly anonymous nobody who writes codes for closed-source mega corp CRUD apps. Being a tech "public figure" (Bryan Cantrill calls it "nerd famous") sounds absolutely awful. Every little thing that you wrote on the Internet in the last 30 years is permanently recorded (!!!), then picked apart by every Tom, Dick, Harry, and Internet rando. My ego could never survive such a beating. And, yet, here we are in 2025, where Ted T'so continues to maintain a small mountain of file system code that makes the Linux world go "brrr".
Hot take: Do you really think you could have done better over a 30 year period? I can only answer for myself: Absolutely fucking not.
I, for one, am deeply thankful for all of Ted's hard work on Linux file systems.
There are plenty of "nerd famous" people who manage it by just not being an asshole. If you're already an asshole being "nerd famous" is going to be rough, yes, but maybe just don't be one?
Only charlatans completely avoid saying things that could get them into trouble.