I hate literally every second of using Google Cloud
October 29, 2025 at 2:44 PM
I hate literally every second of using Google Cloud
I accidentally launched a podcast search app on Product Hunt this morning
it's currently closing in on #9: www.producthunt.com/posts/vol-2
why accidental?
after hitting publish I wanted to polish the svelte app a bit more, but couldn't cancel so yeeted it into the future
then forgot about it...
it's currently closing in on #9: www.producthunt.com/posts/vol-2
why accidental?
after hitting publish I wanted to polish the svelte app a bit more, but couldn't cancel so yeeted it into the future
then forgot about it...
September 22, 2025 at 1:08 PM
I accidentally launched a podcast search app on Product Hunt this morning
it's currently closing in on #9: www.producthunt.com/posts/vol-2
why accidental?
after hitting publish I wanted to polish the svelte app a bit more, but couldn't cancel so yeeted it into the future
then forgot about it...
it's currently closing in on #9: www.producthunt.com/posts/vol-2
why accidental?
after hitting publish I wanted to polish the svelte app a bit more, but couldn't cancel so yeeted it into the future
then forgot about it...
I think most of my bsky network is either 'ML researcher' or 'AI sceptic'.
If you're in the latter group you're going to violently disagree with this: mbanerjeepalmer.substack.com/p/should-pro...
If you're in the latter group you're going to violently disagree with this: mbanerjeepalmer.substack.com/p/should-pro...
September 8, 2025 at 9:47 AM
I think most of my bsky network is either 'ML researcher' or 'AI sceptic'.
If you're in the latter group you're going to violently disagree with this: mbanerjeepalmer.substack.com/p/should-pro...
If you're in the latter group you're going to violently disagree with this: mbanerjeepalmer.substack.com/p/should-pro...
Gmail bizarre spam filtering reaches new level as it filters out Google's own culture newsletter
September 1, 2025 at 3:31 PM
Gmail bizarre spam filtering reaches new level as it filters out Google's own culture newsletter
Mightily tempted to build 'study LLM behaviours as a service'.
So many of these scientific papers are literally just evals.
1. We wrote a set of questions
2. We assigned scores to types of responses
3. We asked the questions to a set of LLMs and scored the answers
4. Journal article
So many of these scientific papers are literally just evals.
1. We wrote a set of questions
2. We assigned scores to types of responses
3. We asked the questions to a set of LLMs and scored the answers
4. Journal article
August 29, 2025 at 5:58 PM
Mightily tempted to build 'study LLM behaviours as a service'.
So many of these scientific papers are literally just evals.
1. We wrote a set of questions
2. We assigned scores to types of responses
3. We asked the questions to a set of LLMs and scored the answers
4. Journal article
So many of these scientific papers are literally just evals.
1. We wrote a set of questions
2. We assigned scores to types of responses
3. We asked the questions to a set of LLMs and scored the answers
4. Journal article
2/2 I think Rich's attitude towards LLMs contrasts with the attitude that first got me into Svelte.
LLMs open up programming. The 'robot holding your hand' is a *good* thing for new coders.
Maybe I'm wrong, but I think a positive direction is possible here.
LLMs open up programming. The 'robot holding your hand' is a *good* thing for new coders.
Maybe I'm wrong, but I think a positive direction is possible here.
August 29, 2025 at 5:55 PM
2/2 I think Rich's attitude towards LLMs contrasts with the attitude that first got me into Svelte.
LLMs open up programming. The 'robot holding your hand' is a *good* thing for new coders.
Maybe I'm wrong, but I think a positive direction is possible here.
LLMs open up programming. The 'robot holding your hand' is a *good* thing for new coders.
Maybe I'm wrong, but I think a positive direction is possible here.
Shout out to @logfire.pydantic.dev for being surprisingly helpful.
Certainly lots of rough edges. But compared to other platforms, which seem to be built for an o11y in-group, Logfire says "read the docs? lol we know you cba, copy-paste this instead"
Certainly lots of rough edges. But compared to other platforms, which seem to be built for an o11y in-group, Logfire says "read the docs? lol we know you cba, copy-paste this instead"
August 22, 2025 at 5:03 PM
Shout out to @logfire.pydantic.dev for being surprisingly helpful.
Certainly lots of rough edges. But compared to other platforms, which seem to be built for an o11y in-group, Logfire says "read the docs? lol we know you cba, copy-paste this instead"
Certainly lots of rough edges. But compared to other platforms, which seem to be built for an o11y in-group, Logfire says "read the docs? lol we know you cba, copy-paste this instead"
Shameless plug of what I wrote on the topic digitalfrontier.com/articles/rss...
August 7, 2025 at 9:35 AM
Shameless plug of what I wrote on the topic digitalfrontier.com/articles/rss...
1. To me IPO counts look relatively flat, apart from a spike in the dot com bubble.
2. The data cuts off six years ago. So it cuts off the enormous spike.
3. Companies staying private longer is an important drag.
(Only skimmed the paper, maybe I missed something that explains this.)
2. The data cuts off six years ago. So it cuts off the enormous spike.
3. Companies staying private longer is an important drag.
(Only skimmed the paper, maybe I missed something that explains this.)
July 30, 2025 at 8:44 AM
1. To me IPO counts look relatively flat, apart from a spike in the dot com bubble.
2. The data cuts off six years ago. So it cuts off the enormous spike.
3. Companies staying private longer is an important drag.
(Only skimmed the paper, maybe I missed something that explains this.)
2. The data cuts off six years ago. So it cuts off the enormous spike.
3. Companies staying private longer is an important drag.
(Only skimmed the paper, maybe I missed something that explains this.)
"We have a cleanup on aisle 5. Designer to aisle 5 please.
July 30, 2025 at 8:22 AM
"We have a cleanup on aisle 5. Designer to aisle 5 please.
A fun thing about using LMArena as a daily driver is you get to see *just what an absolute lemon* Llama 4 was on a regular basis.
(And yes, it was, inevitably, CORS.)
(And yes, it was, inevitably, CORS.)
July 25, 2025 at 8:29 PM
A fun thing about using LMArena as a daily driver is you get to see *just what an absolute lemon* Llama 4 was on a regular basis.
(And yes, it was, inevitably, CORS.)
(And yes, it was, inevitably, CORS.)
Both models on LMArena suffering a crisis of self-confidence
July 22, 2025 at 8:50 PM
Both models on LMArena suffering a crisis of self-confidence
Andreas's thing is surprisingly exciting.
Team Blobby.
Team Blobby.
July 21, 2025 at 9:28 PM
Andreas's thing is surprisingly exciting.
Team Blobby.
Team Blobby.
technical diagram of of how gmail decides between '🏷 Promotions' and '🛈 Updates'
July 17, 2025 at 11:51 AM
technical diagram of of how gmail decides between '🏷 Promotions' and '🛈 Updates'
IT'S HERE
YOU KNOW WHAT IT'S THE YEAR OF
YOU KNOW WHAT IT'S THE YEAR OF
July 16, 2025 at 5:18 PM
IT'S HERE
YOU KNOW WHAT IT'S THE YEAR OF
YOU KNOW WHAT IT'S THE YEAR OF
very strange advice from google
July 8, 2025 at 1:02 PM
very strange advice from google
June 16, 2025 at 10:06 AM
coincidentally, yes 4 did just now on @lmarena.bsky.social
June 7, 2025 at 4:12 PM
coincidentally, yes 4 did just now on @lmarena.bsky.social
Product managers: you aren't supposed to put the metric name in the email
June 7, 2025 at 3:52 PM
Product managers: you aren't supposed to put the metric name in the email
IMO this diagram, 'Musk barely follows anyone left wing' would have been sufficient.
If NYT needs to pad it out then make your audience smarter by pointing out the limitations, instead of making them seem clueless overdrawing conclusions.
If NYT needs to pad it out then make your audience smarter by pointing out the limitations, instead of making them seem clueless overdrawing conclusions.
June 3, 2025 at 4:07 AM
IMO this diagram, 'Musk barely follows anyone left wing' would have been sufficient.
If NYT needs to pad it out then make your audience smarter by pointing out the limitations, instead of making them seem clueless overdrawing conclusions.
If NYT needs to pad it out then make your audience smarter by pointing out the limitations, instead of making them seem clueless overdrawing conclusions.
Substack sends too many emails
May 23, 2025 at 2:15 PM
Substack sends too many emails
![[normal Winnie] Scrolling (the other place)
[tuxedo Winnie] Reading Substack
[tuxedo WInnie with monocle] Reading Substack on paper](https://cdn.bsky.app/img/feed_thumbnail/plain/did:plc:aqfi6fxtxbxetvlkvs3wzdtc/bafkreicclt663wkqrvndof32yegtq5rtv36fr5rmmgphaa337ijl2v5o2m@jpeg)
Birthed from the other place: printstack.net/for-readers
- I'll print out your Substacks and post them to you.
- You read them in the park.
- Leave your screens at home.
- I'll print out your Substacks and post them to you.
- You read them in the park.
- Leave your screens at home.
May 6, 2025 at 6:15 PM
Birthed from the other place: printstack.net/for-readers
- I'll print out your Substacks and post them to you.
- You read them in the park.
- Leave your screens at home.
- I'll print out your Substacks and post them to you.
- You read them in the park.
- Leave your screens at home.