Also have you thought about partnering with Elicit? That would be cool

The one about how placebos work even if you know they are a placebo

I keep on thinking about starting a D&D campaign, and then I think about scheduling and I preemptively give up lol

I'd love to see the not-good-enough version too lol
godspeed

you just need to be famous enough that DJT invites you to the white office like Mamdani XD

can i see the demo 👀
holding a high bar for pmf is wise

how much do they cost? I would like one!

Congrats @cazwis.bsky.social ! APWOT is such an inspiration to me (and many others, I'm sure)

I think it’s kind of the tiktokification of everything - preference for algorithms by tech companies that let anyone go viral. So having followers means less over there?

HAWT

nice work!!

ikr?! the bad news is that the world keeps changing but the good news is that the world keeps changing

#10 - Plastics warp, accept it and design accordingly! Most of the time, you can fix this by making sure the plastic parts are fully "captured" (surrounded) by another part, so together they form a dimensionally stable unit. I wasted time fighting this instead of accepting it.

#9 - When debugging issues with a customer, assume that they are right about the problem, but wrong about why it's happening. One time, we spent hours debugging a strange connectivity issue that turned out to be... a typo in the user's wi-fi network 🤦

#8 - When you receive a sample from a supplier, try your best to let go of the hope that it's good, and switch into QA engineer mode. Put it through its paces. You'll feel incentivized to speed through and approve but it won't take long for that you bite you.

#7 - The hardest mechanical issues I've had to debug have always been solved by simplifying the problem - making a jig or a model that isolates the issue. Sometimes the process of making the tools itself reveals the solution.

#6 - For a connected product, different problems can be solved at different layers - plastics, boards, firmware, cloud. Plastics can be simple but slow, firmware can be effective but complicated to roll out. Carefully choose the right hammer.

#5 - I always thought that ID drawings had to be really beautiful, but I've been surprised how effective lo-fi, high-speed doodles and cardboard prototypes have been. The point is to unblock a decision, after all.

#4 - I was surprised how much of PCB schematic design is reading the datasheets and trying to follow the reference implementations closely. If you have been avoiding learning how to read a datasheet... let that end today!

#3 - Some components need to be babied - like a Raspberry pi, we had multiple brownouts and flakey boards with early prototypes. Our ESP-32 boards on the other hand are total workhorses. This was counterintuitive to me because I thought RPi was supposed to be beginner friendly.

#2 - 3D printing has quietly gotten way better over the last 10 years! Many people do not realize when a poetry camera prototype is 3D printed. Some, when learning it is, think it's better quality than generic Amazon plastic stuff, which makes sense. The tolerances are better.

100 things I've learned about hardware by building poetry camera

#1 - Hardware has gotten faster! Rapid aluminum tooling can take ~1 week and create thousands of units, if you're willing to sink in ~10-50k

Met with our supplier for poetry camera! Three cheers for practical applications of mandarin!

watched the new kirby game direct and my inner child is so happy

ryan, kelin, and evan assembling poetry cameras at a table

a half-assembled poetry camera, next to columns of components

wip — poetry.camera special drop for config

see you in sf next week :)

~
pics: ig @/brendanhall
w/ @flomerboy.bsky.social @eka.hn

10 minutes in KiCad (for @poetry.camera)
song is Sunny's Time by Caribou

hand holding a poetry camera against a sky blue background

photo from the smithsonian american art museum's new shape of power exhibit, with a humanoid sculpture against a dark background. The sculpture resembles an overweight man covered entirely in motifs in patterns. It is called "DNA Study Revisited," by Roberto Lugo, 2022

if you're in the DC area & want to play with @poetry.camera

we're partnering with the Smithsonian American Art Museum & Monument Lab on a series of workshops related to their new Shape of Power exhibit

first workshop on thurs march 6. @flomerboy.bsky.social and i will be there!

maybe text blast should look less like a primary CTA? I read it as meaning something like "invite people"

I can type now but now I'm not sure what goes in there. phone numbers? separated by commas?
Oh I guess I also don't know whether I should click the going etc buttons as the doc creator