monroewalker

Thank you for sharing, that provided some good context for how to interpret this

Post content:

_____

I wish we didn’t need these again, but here is the honest version of Anthropic’s biology announcement (Caveat: I haven’t worked in bioinformatics for many years.) The good: Anthropic ran ~950 Claude agents over a large biological sequence database. Claude searched, wrote code, compared sequences and genomic neighborhoods, and found an interesting pattern that apparently had not been noticed before: a known reverse transcriptase associated with another gene and a repetitive DNA array.

That is cool. Automating this kind of open-ended bioinformatics search at scale is useful, and Claude may have found a lead a human would have missed.

But: Claude did not do a biological experiment. It searched databases and analyzed data.

Humans then took the candidate into the wet lab. And the wet-lab result so far is modest: they showed that the repeat array produces short RNAs.

We still don’t know what the system does. No function, mechanism, phenotype, targeting, defense activity, or programmability has been demonstrated.

This is also where the CRISPR framing gets ahead of the result. Right now, “it has some features reminiscent of known programmable systems” is a hypothesis for what to investigate next, not a discovery that it behaves like CRISPR.

And there is a missing baseline: bioinformatics has had tools for finding unusual gene neighborhoods and candidate systems for years. The interesting comparison is 950 Claude agents vs. an expert using the best existing computational pipelines - not Claude vs. someone manually looking through 200,000 sequences.

So my honest announcement would be:

Claude autonomously found an interesting candidate for a previously uncharacterized biological system. A small human wet-lab experiment confirmed that part of the candidate is expressed. We don’t yet know what it does.

That is a good result.

But in a regular biology lab, this isn’t the finished paper. It is the result you show at lab meeting and say: “This looks interesting. Now we need to figure out what the hell it does.”

Maybe that next step leads to a major discovery. But that discovery hasn’t happened yet.

Seems more than good enough for a live model though! I can imagine this demo being extended to be a lot nicer to play with. You can just feed the model engine analysis and it can make as high of quality moves as needed. No longer any correlation between the model's understanding of the position and the moves that would be made but I think that's still a really nice improvement when thinking about this as adding live voice interaction to existing chess vs computer functionality rather than adding chess to possible interactions with the latest live voice model.

Same price as Fable?

Great summary! I would just add that cherry on top though -- that HuggingFace tried using the top commercial models in response but couldn't because of the cybersecurity restrictions so they had to use GLM 5.2 instead

"When we started the log analysis, we first used frontier models behind commercial APIs. This did not work: the analysis requires submitting large volumes of real attack commands, exploit payloads, and C2 artifacts, and these requests were blocked by the providers' safety guardrails, which cannot distinguish an incident responder from an attacker. We ran the forensic analysis instead on GLM 5.2, an open-weight model, on our own infrastructure. This had a second benefit: no attacker data, and none of the credentials it referenced, left our environment."

What makes it nicer to use than Slack?

Heh after reading that title card I thought this was going to be a mosquito based software bug analogy. I expected a description of how to write software that resulted in more "good bugs" that might facilitate finding other bugs somehow. Now I'm a little disappointed

Now that Claude 4 is out, I’m making some updates to the project I’ve built primarily just with Claude Code: https://github.com/mwalkerr/BookmarkCanvas

It’s just a basic IntelliJ plugin which provides an infinite canvas to add code bookmarks to. I work on a large code base and often have to take on tasks involving lots of unfamiliar areas of code and components which influence each other only through long chains of indirection. Having a visual space to lay things out, draw connections, and quickly jump back into the code has been really helpful

The canvas and UI is built using Java AWT since that’s what IntelliJ plugins are built on, but it occurred to me that I could just throw in a web view and use any of the existing JS libraries for working on an infinite canvas. React Flow has seemed like the best option with tldraw being what I’d fallback to.

But then.. if the canvas is built with web technology then there’s no reason to keep it just within an IntelliJ plugin vs just a standalone web app with the ability to contain generic content that might open files in IntelliJ or any other editor. I’m pretty sure the “knowledge database on a canvas” thing has been done a number of times already so I want to also see if there are existing open source projects that it’d be easy enough to just add a special node type to

See also this recent post about Mercury-Coder from Inception Labs. There's a "diffusion effect" toggle for their chat interface but I have no idea if that's an accurate representation of the model's diffusion process or just some randomly generated characters showing what the diffusion process looks like

https://news.ycombinator.com/item?id=43187518

https://www.inceptionlabs.ai/news

Just about every line of code for this was written by Claude 3.7 via Claude Code. I never gave the other AI development tools like Cursor and Aider a fair shake so I may have just been behind the times on what's possible with agentic editors, but I found Claude Code to be extremely impressive. The API costs did end up being something around $15 though and this is a really small project, so I imagine cost goes up quite a bit to do any non-trivial work on a project of a decent size

This was great for me though because this is something I've had in mind for a while but never considered quite useful enough to be worth the time it'd take to learn about IntelliJ plugin development + the Java AWT knowledge required to create a canvas app like this

lol yeah I tried to get it to whisper too. And talk faster or slower or do accents. It seemed to be able to kind of do each of those things but only very slightly. Enough to see that there was some successful interpretation of the request but lack of flexibility to fully execute on it. OpenAI's model still has this beat on that front imo (talking quietly / slower / faster)

Yeah after a few interactions, the repetition of the mannerisms that initially added to the sense of life-likeness started to break the illusion a bit. The "you got me" response shows up a bit too often. The creativity remains impressive though

This was already posted here: https://news.ycombinator.com/item?id=43221377 but I’m really surprised at the lack of attention this model is getting. The responsiveness and apparent personality are pretty mind blowing. It’s similar to what OpenAI had initially demoed for advanced voice mode, at least for the voice conversation portion.

The demo interactions are recorded, which is mentioned in their disclaimer under the demo UI. What isn't mentioned though is that they include past conversations in the context for the model on future interactions. It was pretty surprising to be greeted with something like "welcome back" and the model being able to reference what was said in previous interactions. The full disclaimer on the page for the demo is:

" 1. Microphone permission is required. 2. Calls are recorded for quality review but not used for ML training and are deleted within 30 days. 3. By using this demo, you are agreeing to our "

edit: Actually this has been posted quite a few times already and had good visibility a couple days ago: - https://news.ycombinator.com/item?id=43200400 Others: https://hn.algolia.com/?q=sesame.com

Do the banks offer email notifications for transactions? That could be another approach if you automate pulling info from the emails

A cross-platform clipboard manager / search-and-filter tool / launcher built with Flutter that has a simple Python plugin interface.

Plugins can be used to add new "result actions" and new sources of entries to filter and select. Eg. recent Jira tickets, email inbox, shell history, Notion pages, etc. The result actions are a way to easily perform common transformations on selected entries (eg. wrap in triple backticks, find and parse json, trim whitespace, ...) or kickoff some script with a selected entry as an argument.

Project started as a result of having to do a lot of work using Ubuntu and sorely missing Alfred and all the workflows I'd built with it. I wanted something for which I could build workflows once and have those workflows available on whatever system I'm on. Plus to be able to build some plugins that would be usable by coworkers regardless of what operating system they're using and with minimal runtime resource usage. There are some existing cross-platform solutions which could serve this purpose, like Cerebro, Ueli, Script Kit, some others.., but I wanted something lighter weight than is possible with an Electron app. Granted the current state of Epte is that it's built with Flutter + Go + Python so the final distributable and runtime memory usage are higher than is ideal.

Basic Windows support is almost there but there doesn't seem to be a great solution to switching to existing windows of an application instead of just re-launching it. The tool isn't intended to be as good or better than any given OS's built-in launcher so I'll probably just leave that as-is and upload the current state of the Windows build.

https://github.com/mwalkerr/Epte

The best approach I’ve found so far is to just have a single master “event log” where I dump everything that I want to save by default. I have specific places to put things but if I can’t be bothered to decide where or am not sure it’ll just go to the event log. I’m using Notion for this where each entry is its own page in a “database” list. Adding a new page is trivial though through the site or app. I have an iOS shortcut setup too to open the entry creation

Oh that would be awesome for that to work. Thanks for sharing

Anyone else have their zsh history get truncated occasionally? Not sure where in my setup the issue is but sometimes the history file will just be recreated. I have all the histsize and other env vars set correctly and I haven’t been able to correlate it to anything in particular I’m doing when it happens

If servo (or something like it) succeeds, would that mean potentially being able to swap out chromium in Electron? Would that help with performance / application size?

Nice! Will be trying this over the next few days

I've been working on a cross-platform clipboard manager launcher thing and using xdotool to simulate a ctrl+v on Ubuntu. On every paste though, the entire screen would freeze for a solid second. After a lot of exploring options I came across https://github.com/robertwayne/libmutter-fix/tree/main.

The TLDR as I understand it is that in Gnome's Mutter, when a keyboard event is received from a different source than the last keyboard event, some expensive operation is kicked off (`keymap_changed`) which is unnecessary for most cases but required to handle something like switching to a keyboard with a different layout than the one last used.

The Github repo linked above has more info and links to related issue trackers and discussions. The solution implemented in the repo is to checkout libmutter and just remove the line causing `keymap_changed` to be called for `XkbNewKeyboardNotify` events.

Sharing in case anyone else happens to have been banging their head against the wall about this

This Reddit comment mentioned this site with used servers:

https://pcserverandparts.com/ https://www.reddit.com/r/LocalLLaMA/comments/1i8rujw/comment...

I've been pretty happy with Ranger (https://github.com/ranger/ranger)

It’d be cool to have a gutter on the left with the html tag that appears for that line

If building a desktop app with the intention of having JavaScript plugins, would it not be viable to include bun in the app bundle to use to run the plugins? Curious if anyone has played around with this idea