Thanks @HamRadioVillage for such an awesome job organizing and letting me one of the two zero to extra’s this year!

@shadcn @JohnPhamous

Should I finish this and send a pull request? @shadcn @JohnPhamous #designengineering

First try failed epically but can't upload here I guess because the clip is less than a second of video. The print failed and created an alien landscape instead of a flat plate.

The online 3d model for the computer's board is missing the nvme headsink plate. To make a case, it was necessary to precisely measure the screw holes and print a test plate to see if everything lines up: it worked! On the second try.

Not too bad for a version 1. Took 5hrs to print, $1 of filament.

Received my #LattePandaSigma single board computer yesterday but no cases are available yet. Designed one real quick and printed it overnight. Came out just in time to test it in the field with my monocle display.

Sneak peek at my latest project :)

Market Caps: 999 Club: 1K items * 34.8Ξ floor = 34,800Ξ 10K Club: 10K items * 2.795Ξ floor = 27,950Ξ Based on scales & trend data, I expect that: * 10K market cap will become 9X bigger than 999's * 999's floor will be 3X of 10K's #ens market is 🔥

Still, there's more left to do. For 'ezhomeloan', 'ez' should be 1 word or 2 ('e', 'z')? Also, this mostly works for English words but sometimes works for Spanish, French, and similar but not Asian langs like Thai yet.

Additionally, there are nearly a dozen edge cases to handle: * '-' as a separator * pluralization * 'i' as prefix e.g. iPhone * so many more...

The lang detection library I was using doesn't correctly detect 2-3 words, it needs a full sentence. So for compound words, it uses the ML system only. lang lib(as 1 word): German lang lib (as 2 words): Indonesian ml (as 2 words): English IDK why people use the lang library

The first step is to recursively mask the string: marinehunter => ['marine', '%hunt', '%er'] %hunt%er => ["hunter"] => ['marine', 'hunter']

Unintuitively, this was much harder to implement than expected.

9/5/2022 Update: Compound Words! Compound word #ens are now better calculated, accounted for, and used for better language detection.

Thanks @JPEGvaultDAO @artsmandoteth for being amazing hosts and having me on for this very personal conversation.

The updated system uses an ensemble machine learning (ML) algorithm where several identifier functions are used in parallel, and the results are compared and decided via ML. Previous accuracy: 36% New acc: ~75% [1] No increase in queue times.

Needed this. DM me if you think I should add a specific language. Once enough appraisal refreshes are done, it should be able to weigh non-English words the same way it is doing English words (concreteness, animacy, frequency, etc).

The power of better language models: Left is esti bot comparables for "van. com"; right is @ENSAppraiser for "van.eth". If only the length was factored in, 1e2.eth, triple emoji, or triple digits would be listed. But it's not the only factor, here. #ens #ensdomains

I wanted there to be a non-pay-to-win way to do this so this feature will stay.