each blockchain selects technical + cultural tradeoffs that make it better suited to some use cases and not others the reason I'm an ethereum fan = it selected better tradeoffs than other blockchains for building The Future of Finance. that's the usecase I find most interesting

9/ @Fundstrat posts bullish thread on $ETH

everyone is forming an opinion about glamsterdam EIP’s, but almost no one understands the technical implications of each proposal here is a direct comparison of the headliner EIP’s along the dimensions that matter:

installing dev tools is now a mostly solved problem I still remember the bad old days of fighting compilers/libs/drivers/envvars for hours just to install basic stuff these days you can robustly install almost any tool using a single call to brew, cargo, npm, or uv

excited for one of my favorite crypto data ppl to join the Paradigm team let me tell you, sybillers freaking hate this guy

fyi rust is super fast in jupyter noteboks now last time I tried jupyter rust a couple years ago, it was unusably slow. 10 seconds or more to run a simple cell. now cells can run in just 100s of milliseconds huge props to evcxr authors and everyone else that made this possible

very thoughtful post from vitalik on copyleft licenses in OSS however I can't say I find the arguments convincing. they make sense in theory, but they really need more empirical evidence because there are also plenty of decent theoretical arguments that go in the opposite direction e.g. copyleft reduces the total amount or total proportion of OSS code for reasons X+Y+Z. or copyleft creates reduces idea diffusion because it is too far from the optimal coordination sweet spot when opposing theories battle, data decides the winner

I love uv so much. setting up a new laptop right now. *zero* use of pip. everything is completely squeaky clean for the first time ever, my jupyter env is completely reproducible, not fucked up or weird at all I can finally treat python envs as cattle instead of pets

amazing work by the duckdb team. ducklake v0.2 fixed one of my two big complaints about v0.1 files are not longer stored in one giant global directory. now there are subdirs for each schema/table my other complaint still applies. filenames are just hashes and not human readable at all. if this is improved, I'd use ducklake for everything!

it's dumb how macOS doesn't let you quit Finder. luckily you can fix this with a simple command: defaults write .finder QuitMenuItem -bool true; killall Finder this will enable cmd+q to kill Finder. it will also stop Finder from crowding the cmd+tab view