Mateo
12.5K posts





A small ship I love: We made Claude.ai and our desktop apps meaningful faster this week. We moved our architecture from SSR to a static @vite_js & @tan_stack router setup that we can serve straight from workers at the edge. Time to first byte is down 65% at p75, prompts show up 50% sooner, navigation is snappier. We're not done (not even close!) but we care and we'll keep chipping away. Aiming to make Claude a little better every day.




Heading 4 is finally here 😤 The years of “just bold the text and pretend” are over. Rolling out now.



Let's not overcorrect with the trad-logo hype. Logos are difficult to analyze, in that the only thing we can say is the following—The purpose of a logomark is to distinguish a brand from the rest. There are a few exceptions, maybe if you're a corn trader, you'd want the logo to be similar to other corn traders and unlike other wheat traders. But by and large, its purpose is to be distinctive and provide a basis for a company's identity. The problem with trad logomarks is the semiological complexity—takes longer to recognize, dilutes distinctiveness and they become a fuzzy blob in the visual field, especially if they're in a high visual noise environment with other logos (retail shops, catalogs, etc). One of the culprits is the outline shape: shields, roundels, plates, etc. are SNR destroyers. Strong ("minimal") logos have high SNR and one of the main reasons for that is a distinctive outline shape. Best logo designers achieve a great balance between visual complexity and recognizability. Lance Wyman, Anton Stankowski, Paul Rand, Ikko Tanaka, Ivan Chermayeff, Kashiwa Sato and many others knew this. No one puts this so explicitly, and in a highly impactful presentation as Saul Bass's AT&T pitch (search on YT). There is a bias in the image below (well known brands vs. unknown logos that I found from victorian era), but try to ignore it and it is obvious, objectively, which side has higher SNR.


Weak ratio. Sad! The navelgazers must be distracted?








