mikeocool

All of these things are perfectly possible with a plain REST API with an Open API spec and using some standard auth options, and an AI client that implements a reasonable “make api request tool” (just like the AI clients implement MCP today).

I think the real value of MCP is that it allowed companies to say “we’re doing AI!” When they built an MCP server. Just saying “use our api” was a lot less exciting.

Giving it a different name probably also helped cut through politics at companies where non-technical people didn’t want to open up user data with an API, but they did want to do AI.

I believe you can use your codex subscription with Pi, and it’s officially endorsed by OpenAI and uses your subscription usage.

Though for cc you are correct, using it with your sub draws from extra usage.

Neovim’s current donation page says that funds are managed by OpenCollective — a 501c6 organization in the US. 501c6’s are tax exempt.

So assuming they are the holders of the bitcoin as well, there would be no tax liability.

Previously having a deck was basically a signal that said “I took some time to prepare for this meeting and put together some coherent talking points” even if the deck itself was just a bunch of of text outlining what the speaker was saying.

Now that people are generating decks with AI, they’re basically worthless. I don’t need a bunch of bullet points you didn’t take anytime to think about on the screen while you talk about something you didn’t prepare for.

One person can build a $22 billion company (in year 2000 dollars) that sells an operating system used by 97% of personal computers, the world’s most popular office suite, along with a bunch of other software used by virtually every company in the world, and is on the verge of releasing a wildly popular gaming console system?

Do you have any examples of this?

JOSM is great when you get really into, but if it's really your first edit, just use the editor on openstreetmap.org.

If it's something you end up enjoying, one really satisfying thing to do is pick a small rural town and improve the map. With a few hours, it's not hard to make the OSM data more detailed and better than any other publicly available map.

Given this feature seems to be premised off guessing a user’s age based on their inputs, and a common use case for Claude it uploading personal legal and tax documents to have Claude explain them to you, I am guessing Anthropic is not hurting for data about their customers.

Seems more likely this is designed to stave off similar laws /regulations to what the social media platforms are now being hit with, and moving liability from Anthropic to parents. If dad lets kid use his Claude account for a school project, and Claude ends up telling the kid to hurt himself, it’s now dad’s fault.

Have you never had a conversation someone you trust that you prefer there not be an automated written record of?

It also summarizes your full conversations.

People have plenty of conversations where if one party intentionally chose to record it, it would be big violation of trust.

This is basically just "always on" recording/transcribing that people have to intentionally remember to turn off, that's a pretty big difference.

Yeah, even more weird was when they highlighted the privacy aspect of it. Sure, it maintains the wearer's privacy, just not that of anyone around them.

> They're packing lots of signal into fewer words

“The load-bearing seam is real” or “Autumn hits different” appear to have absolutely no signal in them.

The President of the United Stated doesn’t actually have the authority to tell anyone besides the executive branch what to name geographic places. So the order only forces official US government maps created by USGS and the like to be updated. Other maps can choose to update if they want to.

Open Street Map tends to reserve the “name” tag for what things are commonly called by the local community. It appears that for the Lake Ontario and The Gulf of Mexico they’ve added an “official_name:US” tag to capture the names in the executive orders.

Companies got to release an MCP server for their product and tell their investors they were pivoting to be AI native.

Kinda sounds like you are finally realizing the dream of XML and DTD envisioned 25 years ago.

Lots of standards are terrible without the actual tools on top of them (try making an HTTP request with just a network socket).

But that doesn’t mean we should have single centralized tool that does everything.

I’d posit there’s no big loss if you broke these things up and each had their own copy the repo (if needed):

- CI/CD

- issues

- code review/pull requests

- release artifact distribution

- discovery+social credit for open source

You could host your repo on any dumb storage somewhere that speaks git+ssh or git+https and each tool could pull from there.

> You need centralization (or at the very least, a standard) for that.

We have a standard! It’s this great decentralized VCS called git. For most of the stuff integrations do, they’re just showing code or commits — both of which are readily available in a copy of the repo.

Probably the one thing that’s missing is a standardized webhook system so that your git repo storage could tell integrations when stuff changes. Though even that isn’t strictly necessary if the integrations keep a copy of the repo up to date.

There’s still a world difference between communicating with Claude and managing humans (well).

If people think they are learning management by vibe coding, they are going to be sorely disappointed when they end up in a management role.

If it is like management, it’s like the most low effort version of management.

You just blindly tell it what to do without any regard for its motivations or morale. If it does something wrong, you just delete it and slightly rephrase your instructions and have it try again.

Guess it was a bad idea for everyone to switch to a browser made by one of the world’s biggest advertising companies.

As far as I've seen /clear is the same thing as starting a new session.

If you type /resume right after clear, the first thing in the list is the session you just cleared.

I'm sure that's technically true -- but as long as your personal item doesn't have wheels they seem completely unconcerned with the size. I frequently travel with a duffel bag carry-on that goes in the overhead. On most flights they let me waltz onto the plane with it and a backpack long after they start forcing everyone to gate check their rollerboards.

They could also just say "only one carry-on can go in the overhead bin."

My favorite but of airport-ese that emerged post-9/11 is that on most US airlines you can now have 1 carry-on and 1 personal item.

As far as I can tell the difference between them is that a personal item can't have wheels.

Oh, that’s good, because if you automate something then you’re automatically absolved from all responsibility for its actions.

And you certainly shouldn’t bother monitoring what it’s doing.

> appear to be profiting from is "an invitation-only programme that lets you earn money from the performance of your eligible public reels, photos, stories and text posts."

The difference between “We ask white nationalists to join our invite only pay-for-views program” and “we commission white nationalists to make content” kinda seems like splitting hairs.

Yeah, by most definitions at least 50% of “40-somethings” are solidly millennial.

I think AI is a great learning tool - I have learned a ton from it.

But when I just ask to do a thing and then read the code (typical in a professional environment when there’s pressure to get shit delivered), i have not found that be an effective learning method.

When I’m learning from it I typically go out of my way to have it explain how to do things and things are the way they are. And I often leave the actual typing the code to myself, since that tends to force an active understanding that’s not required when you’re just reading.

If folks have other ideas on how to get that active understanding, I’d love to hear them. Because telling people they should type code seems like it’s going the way of telling people to write cursive.

Pre-AI juniors were valuable because on many tasks, it was faster for me (senior eng) to write a quick spec have a conversation with a junior, and have them go off for a few hours or a few days and write the code, and then come back with it for review. I could do that with several junior engineers and pretty reasonably paralellize the work.

Now a junior takes my spec, drops it into Claude, and submits a PR a few minutes later. So I'm back to being the bottleneck -- there's constant pressure to provide specs and review code, and ultimately that process is just me having an indirect conversation with Claude (em dash is mine).

The junior is probably providing negative value, since it would be more efficient for me just to talk to Claude, and they're not learning anything, because it's really hard to learn anything by skimming code that's being pooped out by Claude.

I'm not sure what the solution to this is, I still think we need to train junior engineers. I think my best advice to juniors right now is stop using Claude so much. Use it to plan and answer questions, but you should still be writing code even it's slower. Because that's the only way they're going to learn enough to effectively guide AI and review output and get past being a net negative.

> mentioned a Telegram bot idea

> Fast forward: she now runs Arch Linux with Hyprland (I use Xorg/i3 though), fully riced with cats.

Did she end up building the bot or did the LLM just lead her down a desktop Linux rabbit hole?

This can sometimes be a challenge with OSM, if the way you want to structure data doesn’t align with the way the community has decided to structure data.

I was looking into building a canoeing map based on OSM data. An important thing to map is “is a particular stream between two lakes navigable by canoe?” OSM has a defined canoe=yes/no tag, but it’s supposed to be used for legal access, not navigability.

In the place I wanted to map, canoeing is basally legal everywhere, but there are many streams that no one ever canoes down, because it would be an enormous pain. There isn’t a way to represent this is OSM today, and community tends to be resistant to adding tags that map “subjective” data.

The solution would probably be for me to keep my own database of navigable streams with references to the OSM features, except that OSM features don’t really have stable ids. If another mapper comes along and decides to map an existing stream in more detail, they might choose to delete an existing stream feature and replace it, split it up into multiple features, or join several existing features together — all of which will result in id changes.