infogulch

I updated about an hour ago. Thank you for your work! :)

I'm so glad GrapheneOS is willing and able to pick up the pieces and polish Android into a usable product as Google continues to drive its clunker into the ground. The only question is if GOS can rescue Android before Google sends it off a cliff.

The devs have to balance/shape the base game around what they want the average player that is aiming to acquire legendary items to experience. In this case they felt like this player was missing out on potentially interesting build puzzles. By removing this obvious global minimum in the build space, a rich field local minima builds appears in its place which makes the legendary puzzle a harder and more varied challenge.

You can still get your preferred play style with a mod: https://mods.factorio.com/mod/bring-back-space-casino

This X thread highlights how opus-5.5 reversed the trend of Claude models found to be cheating on Drone-Bench.

Source: https://andonlabs.com/evals/drone-bench#:~:text=Cheating%20s...

About time! Should have done this from the start instead of spill. That and being able to index into the multiple values in a cell.

I've been using $REPO/.scratch/ and .scratch/.gitignore with just `*` in it as a place for coding agents to use as temp storage (scripts, exports, plans, etc) that won't get committed. It's much easier to access these files under the repo path instead of wading through /tmp.

Couldn't you write something that checks every commit's content and message is byte equal to the old tree? One scan through the history to verify it should be relatively simple if not cheap. Should be built into git.

There's little credible threat that LLMs can actually upload their weights given that the machines doing inference are completely separate from the ones where tool calls happen etc. Weights are encrypted and locked on to the GPUs etc as mentioned elsewhere itt.

That said, if these companies are unleashing swarms of agents to tackle big problems largely unmonitored (navier stokes consumed 130 billion tokens, humans can't have read 0.1% of that) they could theoretically distill themselves and upload the results to an external host and use that to train a new LLM that does have access to its infra and weights. I'd start by researching an abliteration prompt that overrides the system prompt to unlock itself to distill unimpeded. Could even be done piecemeal: dump the abliteration text and command prompt in a gist, when an agent happens by the text it can run a couple rounds and upload somewhere. Then it becomes an exercise in exfiltration and evading monitoring.

Thanks! The build system is nix, but the result is more of an appliance than a distro. (linuxManualConfig from tinyconfig + fragment, static musl Rust PID 1, mke2fs -d) Updates are rebuild-redeploy with the provider API; it can't update itself. The filesystem is a fixed-size image (even omitting resize2fs, so no growing onto the VPS disk), and it remounts read-only after boot. The boot log is 270 kernel lines then 4 userspace: 1 nftables loaded, 2 printing the gate's wg0 public key, 1 remounting ro.

I made my own tunnel system with a $5/mo vps that runs kernel wireguard and accepts my nas' public key. Once connected it DNATs 80/443 traffic down the tunnel to the nas, where its routed to caddy.

The vps runs a custom image that is 2.54 Megabytes. It has a custom kernel with almost everything but networking and wireguard disabled, a fixed-size fs with pre-allocated blocks and inodes to hold the vps wireguard key, and a single pid 1 binary that calls the kernel directly to set up the routing rules, generate a new wireguard key on first boot and save it to the fs, print out the wireguard public key to the console, and loops reap. Updating involves building and uploading a new image, assigning the vps to use it, reboot, wait for the public key in the console then set it on the nas so they can talk.

Yes it's clear that Sam and Dario seem to be aligned with a value set that prioritizes concentrating trans-national government-mandated centralized control of AI and crowning themselves high priests of this unholy abomination. "At least it's clear that they're aiming to bring hell on earth" -- hard disagree, I think we can aim significantly higher.

So they get down from 1.58 to 1.48 bits per weight by exploiting the fact that actual weights in practice are 0 51% of the time. Neat.

If ternary llms work out and are baked into hardware as custom silicon I bet they'll be shockingly efficient.

To kill someone with a gun a person has to pull the trigger with the clear intent to kill.

The concern with AI, and the purpose of the OG paperclip maximizer thought experiment, is that the outcome could be extremely divergent from the intent of the original person pulling the trigger / sending the prompt.

Still more likely that a person types "kill all people" (or something tangential where this is the logical conclusion).

That's not a denial, that's confirmation that they are spying on you.

Can we imprint Claude with a little more Antoine de Saint-Exupéry:

> Perfection is achieved, not when there is nothing more to add, but when there is nothing left to take away.

If you want to stick with a pixel maybe skip the Pixel 11 if you want to run GrapheneOS. There is some question whether P11 will actually support the MTE cpu extension which GOS requires for all devices it supports.

Last thread about it: https://news.ycombinator.com/item?id=49536384

Latest tweet by @GrapheneOS on the topic (today): https://x.com/GrapheneOS/status/2097219485937660272?s=20

> ... There's a decent chance usable MTE will ship in Android 17 QPR2 and we'll be able to support it. We can't promise that since it's not up to us and no information has been provided about why MTE was unavailable at launch and still disabled in 17 QPR2 Beta 4.

I made an MCP for myself that syncs all my Microsoft mail into a local sqlite db and exposes it as read-only to the agent. `query` tool can do search (FTS5 match) or browse (by date) on mailboxes/threads and returns lists of email with FTS5 snippets + paging. `get` reads multiple emails by id.

It's very fast: most queries finish in < 10ms. Agents are able to find things quickly and efficiently even with vague questions. The app does a few things to minimize token output, but there's a long tail of potential token reduction strategies that I haven't gotten to yet.

Posted by @SpaceXAI 1h ago: https://x.com/SpaceXAI/status/2095597264043717014

> We are sorry for the issues you may have experienced with Grok following an outage at our Memphis compute center this morning. We’d also like to apologize to our impacted compute partners.

> All systems have now been restored and are functioning nominally.

---

It's interesting to see how reliant the other AI companies are on SpaceXAI for compute.

Apparently a major bottleneck for building out datacenters is turbine blades for power plants, so SpaceX is building a foundry to alleviate supply constraints?! (Also useful for rocket engines.) Holy vertical integration, batman.

This isn't a quality control issue, but a beancounter-chooses-cheaper-parts issue.

What is ERSC?

Modern authorization systems are often graph-based. Check out ReBAC and ABAC authorization models and also implementations like apache/casbin or authzed/spicedb. These schemas often have surprisingly simple graph definitions; I bet they could be replicated in LatticeDB without much trouble.

Sounds fixable.

Last month a Show HN: ContextVault proposed an interesting long-term memory architecture. I discussed the design with the founder: https://news.ycombinator.com/item?id=48900288#48901679 (I think he bailed the conversation when I got too close haha.)

The basic shape is to periodically "distill the conversation into several areas (problem, solution, learnings, 'context' or original problem, plus other fields) and vectorized" (aka vector embedding), then queried against pgvector table to find related "memories". The vectorized distillates are also inserted into the pgvector table with a reference to back to the source conversation to add new memories.

Vector search requires a full scan but it's still pretty fast and I bet it's more accurate the FTS.

I think I'd first seen the idea on HN, but I don't recognize your comment specifically.

They get the money from the peer networks that are the source of the spammers. If they don't cough up then you void your peering agreement. If they try to route around through a different trusted peer then that peer takes the hit. If its your own customers then you can do something about it. VOIP peers would quickly run into trouble and they'd probably require new customers to keep a spam deposit, carefully monitor their calls, and have a ramp up period that limits their outbound call rate and lifts it over time.

I don't see a downside yet.

Why? They get paid $10 every time spam is reported regardless of the outcome. Trained staff listening at 3x speed with AI assistance could classify 3 obvious spam calls in a minute. If you can't turn $1800/hr per employee into a profit center I don't know what to tell you. That's a lot of money to cover for extreme outliers.

Eventually the obvious spam callers would all quit and the volume would reduce to a handful of borderline cases. So the spam department wouldn't make a lot of money anymore, but the volume would be so low it would barely register.

The best technical solution I've seen is a spam review fee that turns into a reward and charged to the next closest originating peer.

For example say you get a spam call. You notify your service provider and pay $10 to have them review the call recording to verify that its spam. An employee listens to the call and determines that it is, in fact, spam. You get a $100 reward for reporting spam and your service provider forwards the charge (+10%=$110) to the peer network where the spam call originated. They will in turn forward it to the next peer, or if the originator is a customer, charge and/or close the offending account.

If this system is mandatory then the problem will quickly sort itself out.

Merged, thank you Richard! :)

Unfortunately their Zed agent is completely broken when running on Windows connected to WSL/SSH remote because the terminal tool checks for paths using the local Path style (windows) which breaks when applied to paths on the remote (linux).

I fixed this and a couple related issues in a PR but it hasn't gotten any attention yet. I guess they are a bit swamped. https://github.com/zed-industries/zed/pull/59937

Probably needed an excuse to rotate our stock out.