Discussed 11 days ago - https://news.ycombinator.com/item?id=48862165 (81 comments)
HN user
gregsadetsky
hi@greg.technology
---
Cofounder of a web mapping startup acquired by Apple.
Received awards from the US Coast Guard and the Governor General of Canada for creating a map that helped rescue 1,700 people during Hurricane Harvey.
---
Currently refurbishing laptops with robots! https://www.reviserobotics.com/ (YC W25).
Also worked on Disco, an open source self-hosting software: https://disco.cloud/
See my talk about it at the Recurse Center: https://youtu.be/z2lP7C8VT6M
And this (slightly less chaotic) Devtools FM episode: https://www.youtube.com/watch?v=Xr3JJpbCRP0
---
And even more stuff that's wildly out of date - https://greg.technology
Shameless plug - a blog post re: LLMs and Mikrotiks from days ago! https://news.ycombinator.com/item?id=48927915
Take a look at adafruit and sparkfun
You can find kits that have a selection of sensors - enough to get started and experiment
My parser is unfortunately not fancy enough to detect optional steps - but it's a great point!
(OP) thanks for this insight, truly - I took another look and ikea.com reviews actually include a whole dedicated "ease of assembly" value! some very easy-looking products do actually rate quite badly on the assembly
adding that column now, should be live in the next 10-20 mins cheers!
the comments on ikea's site seem to agree with you - lots of extremely negative ones about the assembly
I did not do sentiment analysis on the comments (to modify my fudge factor mentioned elsewhere - which is the bulk of my complexity rating computation), but that could be a good next step..!
Feel free to give me some real assembly time values - happy to adjust my very simple formula of steps x number of parts x fudge factor :)
It’s rather a very simple math formula based on the number of steps and parts.
I adjusted the formula by researching online what people reported as the time it took them to build some of the items. ie there’s a linear regression “fudge” factor - but it’s still an extremely simplified “model”, if you can call it that.
Sorry to hear! The products are from the US site, and I’m hiding products where the extraction pipeline couldn’t get all of the information (steps, parts count) reliably.
What are the missing products?
Also see:
- https://bernardgreenberg.com/Subway/
- https://github.com/BernardGreenberg/NXSYS
- and the incredible pages on the nycsubway.org site such as https://www.nycsubway.org/wiki/Subway_Signals:_A_Complete_Gu...
Have you tried poking at the unifi api?
What was your FirstClass server / where was it? Mine was used by a large-ish (2.5k people at peak) Macintosh User Group in Canada.
It was definitely formative - I still have friends from back then, and remember logging on daily.
As you say, the transition to the real Internet and web was painful/slow for them, and obviously being a commercial software made it especially hard. The software and the user group all disappeared around the same time - hard to say which came first.
had the same error - it worked when I switched to desktop safari fyi
Very fun!
I submitted this more 'nonfiction-y' entry:
https://gist.githubusercontent.com/gregsadetsky/08b6cdd56902...
It looks like it's only one entry per person, so I'm playing fair and not submitting this second entry, which encodes a compressed version of the image using a 128 word dictionary (ie each word encodes 7 bits).
The explanation + the dictionary + the encoded image result in 1000 words.
https://gist.github.com/gregsadetsky/97ac9c8efe9c2a9f08c9cfe...
Getting a Spectrum cable modem internet connection in NYC in 2026 is so deeply humiliating.
Dealing with the ridiculously limited upload speed, the outages, the locked router. The 40 minutes it takes on the phone to get it disconnected. Their constant attempts at upselling you cell phone plans and other terrible tech you’d never consider.
Truly, Fios is the most bare minimum. And there are much better options if you can pay commercial rates (stealth.net! Pilot!).
Truly embarrassing and sad.
I remember - or maybe am mis-remembering - that years ago, SD cards used to exist with wifi access. I'm not sure whether you could read the card remotely? Or the card presented itself as a card, but actually streamed your data? But I remember that they existed?
But the thing that struck me even more is that - again, I may be wrong - those cards actually ran Linux? They were super tiny computers?
In a sense, I find it incredible because - is there a parallel world where we'd all be using SD cards as micro computers, and would just have small docks with usb/ethernet? These could have competed with Pis, could be deployed as micro servers..?
Anyway, if someone has real actual information, I'd love to learn more!
Actually - this article [0] seems to imply that this whole micro-world is sorta dead? But wifi SD cards are real and exist? Do they run Linux...??
I'd say yes - it seems like an interesting / worthwhile project, even if only for you to explore new areas/learn..?
Just curious, have you looked at https://browser.engineering/ - and generally speaking, what are your thoughts on the few new browsers appearing these days?
To the larger point, do you know if Rovio did support/pay Eric in any other way than that hoodie?
Angry Birds generated $500M [0], supposedly.
I would also not be surprised if the Rovio developers, designers, testers, etc. who worked on this game did not get a share of that $500M pie - I actually assume they didn't.
But still, you know. Dare I say it - what about "fairness"? :-)
[0] https://gameworldobserver.com/2023/02/28/angry-birds-2-reven...
Whenever I see Box2D mentioned (the library by the same author as Box3D, obviously), I think back to this story from many years ago
https://kotaku.com/this-guy-created-angry-birds-physics-and-...
Could you talk about your setup? I'd love to learn more. Are you using Cubase or something else? Are input/output/backups complicated / do you commit to a floppy disk process?
Do you recommend finding a working Atari? Are they still reliable in your experience?
Thanks!
More information is also available here - https://victoriaritvo.com/blog/hengefinder/
A beautiful project by fellow Recurse Center alums Victoria Ritvo and John Pribyl!
Also see this interview with Quinn from 2000 in MacTech: http://preserve.mactech.com/articles/mactech/Vol.16/16.06/Ju...
Conversation happening at https://news.ycombinator.com/item?id=48672506
Do not miss out on page 46 and further.
For those curious, this issue includes (on page 20) an article from the creators of the Graphing Calculator, whose origin story [0] often appears on HN.
Also see this [1] for the full list of issues.
I am using 100% of all points from atlas obscura -- see [0] - the data is baked in the html and [1] my source which lists all 30k+ points
[0] https://www.atlasobscura.com/articles/all-places-in-the-atla...
[1] https://gregsadetsky.github.io/clustering-is-not-my-favorite...
agreed re: hotels.com - what I see them doing well:
- don't cluster aggressively ie break out the clusters earlier (even if markers end up overlapping) - allow clusters and non-clusters to co-exist on the same zoom level (this is usually the case - except when maps are implemented to have "cluster only" levels and "point only" levels) - show immediately useful information in the point marker (the price!)
It’s not easy for sure and I’m not a designer or ui/ux expert - just been around maps for a while
I agree with you that the point itself could show that it’s special - or - rely on users tapping/clicking it and then show all the info
In a sense, maybe conveying that the point has more than one data point attached to it isn’t critical - as long as you can see everything inside when clicking it!
Feel free to email me, would love to jam/chat about this!
(op here)
I typically deal with dots at the same place as a single dot and, when clicked, open an infowindow/infopanel that will let people see all of the points’ data.
ie you do need to deal with this issue (by default, no mapping library will do it for you) - and most clustering librairies (or at least their default settings) will also not solve this for you.
The ui for what happens when a point (or even a cluster) is clicked but that point contains more than one point’s worth of data is its own separate problem basically
yeah, I glued the jacket together - but no, I didn't get it printed professionally. that's a great thought!