bjornsing

9 ms·

91.
by bjornsing 1y ago
I was primarily referring to EU initiatives. But the national ”innovation system” here in Sweden is pretty much the same. It’s a cultural issue, not an organizational one, IMHO.
92.
by bjornsing 1y ago
These initiatives always end up the same way: bureaucrats distributing taxpayer money to the socially focused careerists at the top of large companies and academic institutions (perhaps after they’ve left and founded a startup). It’s just s
93.
by bjornsing 1y ago
Isn’t that pretty much exactly what the OP is saying? He just calls it ”push” and ”pull” instead. Different words, same concepts.
94.
by bjornsing 1y ago
The lines blur though when you start keeping state between batches, and a lot of batch processing ends up requiring that (joins, deduplication, etc).
95.
by bjornsing 1y ago
Why are client certs not more popular? They seem like a pretty clean abstraction compared to all the API keys and JWT tokens floating around.
96.
by bjornsing 1y ago
Yeah that was kind of my feeling too while skimming through this: ”Good luck with that…” It’s not a knowledge problem. It’s a vales and incentives problem.
97.
by bjornsing 1y ago
There are of course statistical methods designed to support early stopping. But I don’t think you can use a regular p-test every day and decide to stop if p < 0.05. That’s something else.
98.
by bjornsing 1y ago
Yeah, sorry. I was thinking of a single hash function bloom filter, which is of course the exception. My bad.
99.
by bjornsing 1y ago
I was just chatting with Claude and it suddenly spit out the text below, right in the chat, just after using the search tool. So I'd say the "system prompt" is probably even longer. <automated_reminder_from_anthropic>Cl
100.
by bjornsing 1y ago
> so that you essentially only pay the cost of a single random read for the entire lookup Why would you ever pay more than that for a bloom filter lookup? I mean, I don’t see how that has anything to do with parallelism in memory subsyst
101.
by bjornsing 1y ago
I think I agree. There’s a lot of utility in a single LLM that can talk to many SaaS and integrate them. Feels like a better path forward than a separate LLM inside every SaaS.
102.
by bjornsing 1y ago
The strategic business dynamic here is very interesting. We used to have "GPT-wrapper SaaS". I guess what we're about to see now is the opposite: "SaaS/MCP-wrapper GPTs".
103.
by bjornsing 1y ago
I’m not sure it’s fair to say Zuckerberg’s only advantage over the Winklevoss brothers was his engineering skills. I think he quickly understood the whole dynamics of that potential business and had a more clear vision. Being able to digest
104.
by bjornsing 1y ago
What used to be cutting edge research not so long ago is now a fun hobby project. I love it.
105.
by bjornsing 1y ago
NeRF is a more complex and constrained approach, based on a kind of ray tracing. But results are obviously similar.
106.
by bjornsing 1y ago
> if you multiplied everything - hyperparameters, initialized weights, training data, etc in a network by 10^6 things will still work more or less the same since the upper range is hardly used (with the possible exception of some small n
107.
by bjornsing 1y ago
> Good point. Could have put a bit more work into that. On second thought, and after looking at my cost estimates, the reason DynamoDB ends up costing about the same as S3 for this kind of use case is storage costs. DynamoDB is a lot che
108.
by bjornsing 1y ago
Yeah, it’s a bit of a chicken and egg problem. Since I don’t have a way to find potential customers I feel it’s too risky investing in stuff like client libraries and good API docs. But I can definitely understand you’d like to see more.
109.
by bjornsing 1y ago
Thanks. Interesting read, and an interesting product / service. Have been thinking about the same approach myself…
110.
by bjornsing 1y ago
> what can you do that redis can't? Keep the data in S3 for 0.023 USD per GB-month. If you have a billion keys that can be useful. > I'm also skeptical of the graph on your front page that claims S3 cost as much as DynamoDB.
111.
by bjornsing 1y ago
Thanks for the honest assessment!
112.
by bjornsing 1y ago
> Key-centric access: instead of partition-based access, efficient access and replay of all the messages with one and the same key would be desirable. I’ve been working on a datastore that’s perfect for this [1], but I’m getting very lit
113.
by bjornsing 1y ago
The weird thing driving this thinking is that cross-AZ network data transfer between EC2 instances on AWS is more expensive than shuffling the same data through S3 (which has free data transfer to/from EC2). It’s just stupid, but that’
114.
by bjornsing 1y ago
I think the problem is that if you want quick access to all messages with a particular key then you have to maintain some kind of index over all persisted messages. So n would be total number of persisted messages as I read it, which can be
115.
by bjornsing 1y ago
Hehe. But I must say HN in general is significantly higher pace than these YC videos, intellectually.
116.
by bjornsing 1y ago
Why do I feel like YC videos are targeted at really slow people? The combination of discussion in slowmotion and exaggerated gestures reminds me of elementary school. I’m sure there are valuable ideas in there, but I just don’t have the pat
117.
by bjornsing 1y ago
Do we know they didn’t just prune problematic bars from the dataset until they found a one with a solution?
118.
by bjornsing 1y ago
Would be nice if they could briefly describe the algorithm. Sounds like they’ve turned the TSP into an integer linear program that they can do branch and bound on, but I’m not sure.
119.
by bjornsing 1y ago
> Why is it hard for a new party to win a seat? There are thresholds to avoid too many small parties / independents getting elected. You need to win 4% of the vote nationally or 12% regionally to get a single seat, and if you do the
120.
by bjornsing 1y ago
No you can’t really vote for an independent MP in the Swedish voting system. They would have to register a party, it’s very hard to win a seat, and there is no guarantee that they will not win two or more seats – thus perpetuating the colle

More ›