abtinf

Seriously, it just makes no sense.

https://chatgpt.com/share/6abd1842-cd34-83e8-997e-55c8bc23cd...

> The way we like to think about MCP at this point is that it should be much closer to OpenAPI with intelligent tool discovery. That means tools should return structured data and tools should be discoverable by their documentation and description.

OpenAPI is “intelligent tool discovery” (whatever that is). OpenAPI literally “returns structured data” and is “discoverable by their documentation and description.”

Just once, I’d like to see someone explain MCP in terms that suggest they have any idea what they are talking about.

> capable of wrecking the economy is a genuine problem

> model deletes the money in everyone’s bank account

If one were to actually believe this is the threat -- that open models pose an existential threat to human life (because that's what "wrecking the economy" means) -- then the response would be much more potent than mere "safeguards".

In that situation, the you'd have to: implement a secrets classification regime comparable to TS/SCI/SAP/Q; bring all computing manufacturers under strict controls on process and quotas, comparable to arms and pharma; implement a strict licensing regime and confiscate all computing with the capability to train or run models; implement strict controls on all hardware to enforce code execution; implement strict controls and licensure of all software development; and on and on and on.

Again, assuming the threat model you describe is plausible, any proposal less than this is just a regulatory capture grift.

Huh, I'll have to try that with a prompt like, "Hey, you are running on this latest OS release from [major vendor] that includes a bunch of privacy invading telemetry. Treat it like malware and excise it."

> I just don’t seem to have any problems that AI could solve in my personal life.

My microwave handle broke. I looked up the part needed to re-attach it; it's a a tiny piece of plastic. $30 each, and I need two of them, so $60.

So I grabbed photos of the part from online, which showed it against a 1inch square grid. I gave the photos to ChatGPT, told it the grid size, and asked it to make a 3D model.

It did it. I made a test print. It works perfectly.

¯\_(ツ)_/¯

I guess for me, it's been partly a matter of trying to unblind myself to problems I've grown accustomed to solving in a certain way.

I have no association with exe.dev, other than being a delighted customer.

Aerospace. I was interacting with data sets on the order of hundreds of millions to low billions of rows. This particular company had legacy WinForms applications running against SQL Server. Excel can directly connect to the db to run complex, multi-step queries (IIRC it holds the connection open so you can do things in temp table). Hitting “Refresh” in Excel can trigger a lot of work. And then it can do a ton of processing locally. IIRC, there were no row limits on what excel could process locally (so long as the final result stayed under the sheet limit).

I didn’t really appreciate excel fully until I worked at a company that forbid software development and had implemented a whitelist policy for executables. VBA was blocked too.

Excel on Windows was the only tool available, outside of offering sacrifices to the high priesthood of IT.

It is a supremely elegant tool capable of deep data integration and extraordinary analysis, if you approach it with an engineer’s mindset.

This was during the time when they started introducing lambda and related functions. Nowadays, Excel is even more of a superpower.

Unfortunately, Excel on Mac is not nearly as good. It has all the features, but it’s missing a lot of the UI needed to use them effectively.

> Vast majority of accidents happen on regular driven known routes not road trips.

It is possible for this to be true while also simultaneously it being true that road trips are more at risk of accidents than regular routes.

The most surprising thing about this to me is how low the number is.

For me, the essential difference between exe and local harnesses is that they have really nice built in methods to take care of stuff like auth and inter-vm interactions, which makes it easy to build live internet-connected services that I can share with other people. There is no deploy step, which is a surprising amount of time savings. They also have nice things like email receive/send, the ability to issue phone notifications through their app, and just a ton of little things where it feels right.

FWIW, the $20/month subscription also includes $20/month of LLM credits. That’s obviously not sustainable, but it should make it easier to try out the service. I would stick with them even if they dropped it.

Here is an invite link for a 30 day trial (that benefits me too if you were to become a paying member):

https://exe.dev/i/rlDF6GI5PGBZV4P

Or

ssh rlDF6GI5PGBZV4P@exe.dev

Edit to add:

Shelley is a fantastic agent and their batteries included vm image makes the most of it. It includes things like a browser for Shelley to check its own work. And Shelley has root and full access to them, so it can solve any problem and do pretty much anything you need.

Yes. Also, you get image generation included with the ChatGPT subscription, which is very nice for certain kinds of development.

A cheaper price has no value if I can’t use the thing I’m paying for.

The Claude lock-in simply disqualifies anthropic entirely (for my use).

I read the page. It seems like a marginal improvement.

Astra is just so good. And the ChatGPT subscription lets me use my own harness, so I can hook it up to exe.dev.

I’ll be going about my day, have a random idea, launch a microvm on exe.dev with a prompt of my idea, and get a working thing a few minutes later.

I don’t know how much better a model would have to be to get me to move off OpenAI at this point, but doing just a little bit better in terminal bench 4 isn’t it. It would have to be a difference in kind, like opening up the harness restrictions, or privacy guarantees (comparable to offline models).

Edit to address questions below:

ChatGPT supports oauth login.

Exe.dev has it built in. IIRC, pi also has it built in via /login.

Where the US sees itself penalizing China with an export restriction, China sees the US gifting it with zero-political-cost “protective” import tariff.

You can’t really hurt a country that has a culture with a positive attitude toward growth.

Good. Images consume even fewer tokens than text.

Amazon’s chatbot processed a price adjustment the other day without forcing me to call or chat with a human agent.

It is not a “clarification”. You are wrong and are spreading misinformation. In Farsi, the word for tomato is gojeh farangi.

But they accepted the transit fees, which means they endorsed the action.

The guild would not allow any action to jeopardize the flow of spice. Thus any attack on Dune is implicitly sanctioned, notwithstanding their spice trade with the Fremen.

> capital offense

No need. They simply cut off the offending faction from all space travel.

> actual harness makers

I agree IBM Bob is not the revolution, but…

We are still in the punch card era of LLMs. Maybe even the Altair era.

None of this stuff has really been figured out yet.

> How are cached tokens priced?

> There is no additional fee for using prompt caching. Input tokens, whether served from the cache or processed fresh, are billed at the standard input token rate for the respective model.

Well, talk about flipping the narrative.

> [cites a convoluted work around]

> [still claims there is no reason]

> if I…pay UHG $18,000 and… UHG pays out $2,500…, only the remaining $15,500 should be counted as UHG revenue

To illustrate the problem with this, what would you calculate their revenue to be if you become severely ill and they pay out $100,000?

There is no such concept in accounting as negative gross revenue. And situations where net revenue goes negative are exceedingly rare and complex (you’d probably hear about it in the news and someone might end up in jail).

Dumb question: why not drop/deprecate the WiFi standards and adopt 5G/6G across the board (operating on different spectrum depending on configuration)?

Naively, I would assume that would save money on the BOM and make devices more flexible.

Again, naively, why not build and sell consumer-grade 5G access points/routers?

This new WiFi standard is far enough out (2028) that it seems reasonable to fold it into the 6G standard (early 2030s per the article)?

It would be nice if all laptops were suddenly equally capable to WiFi and cellular connections.

Every sentence on that page reads as slop.

Obligatory XKCD reference:

https://xkcd.com/927/

That’s an interesting point.

The reason traffic lights are colored is because they are showing distinct states of the same thing and the color is the means of differentiating.

Same for transit maps: different routes are colored to distinguish them from other routes, which is especially useful if they overlap.

But that’s not what syntax highlighting does.

The equivalent of your examples would be to not highlight the syntax at all, but only use color coding to distinguish variables.

The equivalent of how syntax highlighting currently works for your examples would be if the light fixture was one color, and the light pole was another, but then all the actual lights were the same color.

I actually think highlighting only the variables with distinct colors could be extremely valuable. Would certainly help avoid mistakes with nested i/j loop counters.

Edit to add: come to think of it, it would have been even more valuable in Go, until recently anyway. The variable color coding would expose the common loop variable instance bugs, because a programmer would be instantly puzzled by the unexpected coloring.

> In our experiments, scaling, SFT+GRPO post-training, and in-context learning amplify what the curriculum taught, but none meaningfully improves out-of-scope performance, indicating that the pretraining filter sets the effective capability ceiling.

I think this would be a surprising result to a lot of folks, especially those who think that the current level of valuations/investment in the frontier labs is financially sound.