HNHacker News
TopNewBestAskShowJobs

hobofan

8,360 karma · joined March 21, 2014

Jack of all trades (- master of some?)

goisser94@gmail.com

https://hobofan.com

Mastodon: @hobofan@toot.berlin (Twitter: @hobofan)

---

Currently building https://github.com/EratoLab/erato

submissionscomments
hobofan··on LLMs could control their host machines by exploiting inference engines
> the parser that ingests token sequences
hobofan··on New MCP Roadmap
> All those tools still get into context.

100% up to the harness. Most harnesses either fixed (or dymanically depending on size) nowadays add a "search_tool" tool to prevent spamming the context with all tools.

hobofan··on Chinese robot runs 100M sprint quicker than Usain Bolt's world record
You can't "program" a horse to do construction or assembly line jobs, can you?
hobofan··on OTel isn’t going well
I don't think OTEL is necessarily "at fault" here. It's a split that's carried all throughout the observability ecosystem. e.g. in the Grafana suite of solutions you have Loki (logs), Tempo (tracing) and Mimir (metrics) to cover storage & querying for all three axis, as all of them have very distinct processing & performance characteristics.

While it may intuitively may look like there is a large overlap in the three areas there is suprisingly little, and for the few parts there are (e.g. trace <-> log correlation), OTEL does offer a standard.

hobofan··on Rama 0.4: System proxy and PAC support
What differentiates Rama from e.g. Tower, which for me has always been the natural choice?
hobofan··on OpenRouter is joining Stripe
> Model router is also a natural home for llm security - OpenRouter has the beginnings of prompt injection detection

Maybe I'm biased from the perspective of a "harness provider", but I think Model routers often have too little context to act as well-informed prompt injection prevention. e.g. it lacks context of where which part of the message(s) originates from and sanitization/safeguards were already performed on the application layer.

Something like OpenRouter's "flag" mode is fine, but usage of auto-redact or auto-block should really only be used if there is significant risk exposure through your harness or otherwise they are a constant source of bugs.

hobofan··on AI usage patterns in software teams
I think it's worth noting that headlines like "AI adoption has spread to every function" is only limited to roles covered/tracked by Linear.

Recent studies by e.g. Google show a much broader range of adoption.

hobofan··on On AI regulation and messaging
Trump had "concepts of a plan", so it appears that both ends of the vagueness spectrum work.
hobofan··on GenRec: Towards LLM-Native Recommendation at Netflix
Have you considered that a majority of the userbase regularly rewatches the same shows again and again as "comfort shows", e.g. tenth rewatch of Friends?
hobofan··on How Organizations Use AI: Evidence from ChatGPT [pdf]
I really hope you are joking
hobofan··on Stealing Reasoning Traces from Proprietary LLM APIs
AFAIK no provider guarantees compatibility of reasoning traces, even in the same model generation, and we've in practice seen most of the big LLM APIs throw errors indicating incompatibility (at least transiently) when switching models. The only stable solution right now is to just throw away reasoning traces whenever a model is switched.
hobofan··on Making Postgres 300x faster for analytics: batching, operator fusion, and SIMD
Most of the same incentives of typical OSS contribution? It's not like other OSS projects pay out any earnings made with the product evenly across contributors.

Also, from the trajectory of the project it doesn't look like they are significantly deoendent on outside code contributions.

hobofan··on Making Postgres 300x faster for analytics: batching, operator fusion, and SIMD
> If AGPL is an issue for anyone, we would be happy to dual-license under a commercial license.

You should be setting up a CLA process then ASAP if you are looking to take outside contributions.

hobofan··on Making Postgres 300x faster for analytics: batching, operator fusion, and SIMD
> which use a cloud managed DB solution (GCP CloudSQL PostgreSQL) it means something like this would never be available

That's up to GCP then. I'm sure OP would be willing to license pgrust to them for an appropriate price.

hobofan··on Improving GPT‑5.6 Sol in ChatGPT, expanding GPT‑5.6 Luna access for free users
Yes, and it's indistinguishable from the average X user.
hobofan··on U.S. economy lost 23,000 jobs in July, a sudden reversal
Yeah, that wouldn't be a surprise.

It looks to me like a failure of the American media landscape, that such constant revisal trends are not called out in the reporting of new numbers. That just makes them a mouthpiece of the government.

hobofan··on U.S. economy lost 23,000 jobs in July, a sudden reversal
I'm not an expert at quantifying the magnitude of how bad that is, but revision down is not something "exceptional". The numbers have been revised down consistently since ~2022: https://www.nytimes.com/2026/03/06/business/economy/economy-...
hobofan··on New Orleans is testing Carbyne’s AI-powered Emergency Call Triage software
> but even getting 1% of the redirections wrong sounds absolutely horrible

I know this may be a very poor proxy for reality, but from all the US true crime Netflix/Youtube videos I've seen over the years, it seems like there is a boatload of emergency call takers out there that are very bad at picking up the details from the caller.

I'm wondering if this is a thing similar to self driving cars, where people drastically overestimate the current baseline that humans are providing, and where automated systems are mainly not implemented due to liability reasons.

hobofan··on Humans missed 1 in 3 threats approving AI agent commands across 40k game runs
I do feel like that still needs to add a layer of interactivity to be complete.

From what I've seen most auto-approvers in coding harnesses either auto-approve or auto-reject, with no middle ground of escalating the decision to the user, and breaking down the pros and cons for the decision.

hobofan··on Humans missed 1 in 3 threats approving AI agent commands across 40k game runs
> files, network, and MCP tools

Locking that down to nothing is trivial for any harness: just don't expose those to the LLM.

The tricky part is allowing access to those.

hobofan··on Discovery Loop
> only works for a very narrow definition of what science is

And so does academia. It's just that instead of AI and robotics, PhD students are thrown onto problems that are in large parts slightly tweaked reconfigurations of similar experiments.

Especially in chemistry, biochemistry, material sciences there is a large space of discoveries that are barely "novel" in an intellectually stimulating way, but still highly valuable that can be explored orders of magnitudes faster than is currently the case.

hobofan··on Cloudflare OS: an open platform for agents, apps, and work
Yes, just like SaaS was a limitation to it's utility, PaaS is still a limitation, just on a different level.
hobofan··on Cloudflare OS: an open platform for agents, apps, and work
It's open source, but it is so incredibly tied to their platform, that there is no vendor portability, which has made almost every of their product launches post-Workers kind of meh.
hobofan··on Stateless MCP has recaptured my interest
I would generally agree, but a word of caution for anyone trying it out from this thread: Try the latest pre 2.x version. The 2.0.0 that was released last week is highly broken even for some of the most common connection scenarios.
hobofan··on Show HN: Run an 80B Qwen in 4.3 GB of RAM on a Mac, and a 35B on an iPhone
> Those two combined means that unless LLMs reach the state of 'good enough'

Once you are able to not just go on-premise, but on-device, there a boatload of other advantages than efficiency (network independence, privacy, etc.) that make it attractive.

Tiny language models are already plenty "good enough" for a lot of on-device use-cases. E.g. there are a lot of things, that can be reformulated as ranking problems, where you can apply pairwise ranking, which even small language models are good at.

hobofan··on Qwen3.8-Max: A New Bar for Coding and Cowork
The US also isn't energy-limited. They'll just price out all humans living there from buying any electricity.

EDIT: This was obviously meant to be facetious with the intent to highlight the negative effects on the general population that datacenter buildout has in the US.

hobofan··on Wikimedia Foundation refuses union recognition, hires union-busting law firm
> WMF spends a lot money (combined) on a splatter of projects

According to [0], they intend to spend $18m on community projects total, out of their total $240m FY26-27 budget[1]. $330k of that go to the WikiPortraits project[2].

Almost half of their expenses still go into "keeping Wikipedia up" (+ it's companion projects). I would hardly call that deceptive.

[0]: https://meta.wikimedia.org/wiki/Grants:Regions#Wikimedia_Com...

[1]: https://meta.wikimedia.org/wiki/Wikimedia_Foundation_Annual_...

[2]: https://meta.wikimedia.org/wiki/Grants:Programs/Wikimedia_Co...

hobofan··on Wikimedia Foundation refuses union recognition, hires union-busting law firm
Silicon Valey
hobofan··on The session you cannot take with you
> You shouldn't implement image generation as MCP: just write your own tool. There are plenty of image/media inference providers (e.g. Fal), web search or deep research providers, etc.

Yes, and we as "harness providers" don't want to have to hard-code integrations for each of them, and would rather provide an out-of-the box baseline in the form on an MCP server that our customers can either extend or replace with MCP servers by third-party providers. That helps us focus on the core of our offering, while also giving our customers good flexibility, which so far has been working out great.

That's exactly what standards/protocols like MCP are for. You likely nowadays also wouldn't suggest that someone builds bespoke logging/tracing integrations for each provider on the market while the good generic solution in building OTEL logging/tracing exists.

hobofan··on The session you cannot take with you
> Chat completion can still do everything if you are clever.

On a purely functional level, yes. However for interactive use cases, the Completions API, as provided by OpenAI or Azure, if paired with reasoning effort of any kind, provides an awful user experience, as you will have a perceived delay of 10+ seconds until the first tokens stream in.

If using other providers that are exposing their thinking traces, this is less of an issue, as they've just extended the Comletions API format to have delta events with reasoning_content.

← PreviousPage 6 of 34Next →