tlonny

6 ms·

1.
by tlonny 1mo ago
I much prefer cursor paginated API requests vs. webhooks. The obvious downside being that in order to not get 429'd you need a respectable poll frequency - meaning you lose reactivity to new events. Thus I think webhooks still have a p
2.
by tlonny 2mo ago
As I understand it, anything publicly viewable is free to be scraped without issue. If you scrape stuff behind a sign-up, you potentially open yourself up to being sued for breach of ToS. Now I'm sure the Chinese labs don't give a
3.
by tlonny 2mo ago
I would argue that my proposed flow is far more convenient. I don’t have to sign up, enter payment details and I can interface in plain English.
4.
by tlonny 2mo ago
Not to disparage, but I wrote: "Using the attached image as a stencil/background - please generate a QR code for https://XXX" into ChatGPT and got something valid that looks similar in style to your offering. I thin
5.
by tlonny 3mo ago
Trying to polish a bunch of my projects I've worked on over the years but never had the cojones to release to the wider world: Hallways ( https://hallways.lonnycorp.com ) - a web browser for 3D spaces, where instead of hyperl
6.
by tlonny 4mo ago
I think _some_ but not _too much_ typechecking is the sweet spot for LLMs. Without any typechecking, LLMs obviously find it harder to work agentically and validate their work. With too much typechecking (I'm looking at you, rust), I&#x
7.
by tlonny 5mo ago
Presumably because API keys are n bytes of random data vs. a shitty user-generated password we don’t have to bother using a salt + can use something cheap to compute like SHA256 vs. a multi-round bcrypt-like?
8.
by tlonny 5mo ago
Bugginess in the Claude Code CLI is the reason I switched from Claude Max to Codex Pro. I experienced: - rendering glitches - replaying of old messages - mixing up message origin (as seen here) - generally very sluggish performance Given ho
9.
by tlonny 6mo ago
how does it work?
10.
by tlonny 6mo ago
Indeed - the future is RL meet-ups and small, intimate online communities. Perhaps not the worst thing in the world?
11.
by tlonny 6mo ago
I’d love something like this implemented for email. Sending an unsolicited email to a random person X requires you to pay a small toll (something like 50p). Subsequent emails can then be sent for free - however person X can “revoke” your ac
12.
by tlonny 6mo ago
HN may not be “mainstream” but it is certainly _very_ vulnerable to bot spam given the topics discussed and the make-up of the audience. You can already see it happening now - at least the bots that write like vanilla Claude/ChatGPT. P
13.
by tlonny 7mo ago
Look at his other comments - its textbook LLM slop. Its a fucking tragedy that people are letting their OpenClaws loose on HN but I can't say I'm surprised. I desperately need to find a good network of developers because I think t
14.
by tlonny 8mo ago
Repugnant.
15.
by tlonny 8mo ago
Doors - A first person, exploration game/experience that I built from scratch. Doors lets you explore URL addressable 3D rooms that link together seamlessly via portals. The idea is that people would upload rooms to the internet (to gi
16.
by tlonny 9mo ago
0.0005%*
17.
by tlonny 9mo ago
If this is somewhat common for games, could one create a virtual fs with FUSE that dedupes using via content-defined chunking and install games there? I feel like writes would probably be quite painful, but with game assets are essentially
18.
by tlonny 11mo ago
Also check airline miles haven’t gone missing. A friend of mine recently had his BA account compromised, all his Avios stolen and he was none the wiser after receiving about 60 emails a minute
19.
Ask HN: What am I doing wrong Re Agentic coding
17 points by tlonny 1y ago|16 comments
20.
by tlonny 1y ago
The rank disrespect of somebody asking you to review something they haven't even looked at is eye watering. I feel like AI-induced brain-rot of engineers is inevitable. Unless we see AI leapfrog into something close to AGI in the futur
21.
by tlonny 1y ago
To quote Mean Girls: "Stop trying to make Doxbridge happen".
22.
by tlonny 1y ago
I totally agree. Isaac, if you're reading this - stop sending me PDFs generated by Perplexity!
23.
by tlonny 1y ago
> Gmail is so big that when Outlook, Apple Mail, and even Thunderbird connect to it, they do an OAuth exchange and then talk over a proprietary protocol. Can you elaborate? Anything I can read on this?
24.
Show HN: HydraMQ - Postgres message queue implementation for NodeJs/TypeScript.(github.com)
3 points by tlonny 1y ago|0 comments
25.
by tlonny 1y ago
The product (from the perspective of the consumer) _is_ relatively straightforward IMO. What isn't straightforward is accurately modelling the credit risk of the bonds when this class of debt is securitized (the meat of this article).
26.
by tlonny 1y ago
If AI is what stops me having this exchange with Siri 50% of the time: > Hey Siri, please play Emerald Rush, by Jon Hopkins on Youtube Music. > Sorry, you'll need to continue in Youtube Music. Then yes - I want AI on my phone
27.
by tlonny 2y ago
AI "polishing" tools are essentially a form of anti-compression. Lets take some information represented concisely and needlessly pad it with formalities and waffle so it appears more "professional" (whilst also throwing
28.
by tlonny 2y ago
Given the size of the training data - I don’t think it would economical to validate all training data with high-end LLM models.
29.
by tlonny 2y ago
> For example, CA schools have to publish statistics on suspensions and expulsions. Are there actionable consequences if these numbers get too high? If they're merely published, as a parent, I would see high numbers as a positive si
30.
by tlonny 2y ago
I'm pretty sure you pee it out too.

More ›