It cannot be solved, at least not in the way I think people want it to be.

We’re lonely because we are wired to avoid rejection and uncomfortable social situations, and because technology has given us hundreds of alternatives to sitting in the mess of connecting with people.

You can only solve it in your own life - by being courageous and spending more of your time in the physical world than in the digital one, willing to gro through the shitty feelings that come with being a human trying to meet other humans.

You cannot solve it for other people. There’s no sexy solution here. Meetup.com or whatever dating app or tech platform or not for profit will not fix it, because it takes individuals choosing the hard path and that will never happen en masse.

I loved the first and last third. The middle is meandering and hard to get through. Watching everything unfold though is really satisfying at the end.

Don’t be terrified, it just sounds like you read a lot. Many of these books are huge, most people definitely wouldn’t finish then all in a year.

I do not read a lot. I know people who do. That’s why I’m terrified.

I’d say 12 books a year is minimal, 52 is excellent. If you’re not in that range, you’re in trouble.

That article is more about feasibility rather than desirability. There's even a section where they say:

> Settling the question of whether companies or governments will be ready to invest upwards of tens of billions of dollars in large scale training runs is ultimately outside the scope of this article.

Ilya is saying it's unlikely to be desirable, not that it isn't feasible.

It'll be interesting to see how this goes, but my first impression is that it's actually not where we want to go. One of the cool things about MCP (or even just tool calling) is that the LLM on top of a tool provides a highly flexible and dynamic interface to traditionally static tools.

I love being able to type "make an iptables rule that opens 443" instead of having to dig out the man page and remember how to do that. IMO the next natural extension of this is giving the LLM more capability to generate user interfaces so I can interact with stuff exactly bespoke to my task.

This on the other hand seems the other way round, it's like bolting a static interface onto the LLM, which could defeat the purpose of the LLM interface layer in the first place right?

Giving LLM the ability to generate UI is a cool concept, but our models are not there yet. MCP Apps can be extremely powerful, for example, you can play Doom inside ChatGPT: https://x.com/rauchg/status/1978235161398673553?s=20

I don't think we can generate anywhere close to this kind of UI just yet.

We built https://usefractal.dev/ to make it easier for people to build ChatGPT Apps (they are technically MCP Apps) so I have seen the use cases. Most of these use cases LLM cannot generate the UI on the fly.

That's what we're doing at Hallway (https://hallway.com).

UIs should be fully remix-able and not set by the datasource/SaaS. So we built out a system to allow users to use the standard UI or remix apps as they want. Like Val.town, but with a flexible UX/workspace layer. Come check us out!

> giving the LLM more capability to generate user interfaces

This is not dissimilar to the argument that "MCP needs not exist, just tell llm to run commands and curl". Well, llm can do those, and generate user interfaces. It's just they don't work reliably (maybe ever, depending on how you define "reliable").

I guess as engineers we can do some work and create stopgap solutions or we can all sit and wait for someone else (who? when?) to make AGIs in which everything just magically works, reliably.

MCP has already drastically lost utility already thanks to skills - for most things it is easier to just hand the model a CLI that it can run.

I'd imagine the same thing will happen here: It will prove more flexible to not push the model (and user) towards a UI that may not match what the user is trying to accomplish.

To me this seems like something I categorically don't want unless it is purely advisory.

MCPs as a thin layer over existing APIs has lost utility. Custom MCPs for teams that reduces redundant thinking/token consumption and provides more useful context for the agent and decreases mean time to decision making is where MCPs shine.

Something as simple as correlating a git SHA to a CI build takes 10s of seconds and some number of tokens if Claude is utilizing skills (making API calls to the CI server and GitHub itself). If you have an MCP server that Claude feeds a SHA into and gets back a bespoke, organized payload that adds relevant context to its decision making process (such as a unified view of CI, diffs, et. al), then MCP is a win.

MCP shines as a bespoke context engine and fails as a thin API translation layer, basically. And the beauty/elegance is you can use AI to build these context engines.

In your example, you could achieve a similar outcome with a skill that included a custom command-line tool and a brief description of how to use it.

MCPs are specially well suited for cases that need a permanent instance running alongside the coding agent, for example to handle authentication or some long-lived service that is too cumbersome to launch every time the tool is called.

I mention mean-time to decision making and that's one of the rationales for the mcp. A skill could call a script that does the same thing -- but at that point aren't we just splitting hairs? We are both talking about automated repetitive thinking + actions that the agent takes? And if the skill requires authentication, you have to encode passing that auth into the prompt. MCP servers can just read tokens from the filesystem at call time and don't require thinking at all.

Exactly. The way it’s mostly been used so far is a poor abstraction over stuff you can just put in the context and have the agent run commands.

It really shines in custom implementations coupled to projects. I’ve got a QT desktop app and my mcp server allows the agents to run the app in headless mode, take screenshots, execute code like in Playwright, inspect widget trees, send clicks/text/etc with only six tools and a thousand tokens or so of instructions. Took an hour to build with Claude Code and now it can run acceptance tests before committing them to code end to end tests.

There are important contexts outside of machines you control where installing or running cli commands isn’t possible. In those cases, skills won’t help, but MCP will.

Hence why I said drastically, rather than totally. There are still a few edge cases where it is worthwhile, but they are small and shrinking, especially with services providing UI's with VM's/containers for the model to use increasingly being a thing.

> for free like this

for free: not.

but I'm certain the race has just begun: big service providers and online retailers are currently implementing widgets enabling the purchase of their services and goods directly within the ChatGPT or Claude chat windows.

If you are outwardly meeting lots of people and your therapist is picking up on vibes you aren't awkward, it sounds to me like you might be being quite hard on yourself. Not to suggest your experience isn't valid, but that perhaps your small talk is not the issue!

There's a lot of wisdom in this post and it resonates with my experience, great write up OP.

I'd add one thing though: OP's ability to observe and imitate these kinds of social dynamics he was seeing suggests he's already coming from a solid foundation of EQ and also feeling secure enough to try on these different personas. Often there's a lot of work to be done to even get to that place!

100%. I’ve been neck deep over the past few months in developing a bunch of Windows applications, and it’s convinced me that never deprecating or removing anything in the name of backwards incompatibility is the wrong way. There’s a balance to be struck like anything, but leaving these things around means we continue to pay for them in perpetuity as new vulnerabilities are found or maintenance is required.

ChatGPT can only do this now because the information is essentially freely available. Booking.com etc post their pages on the web to get traffic. In the world OpenAI is imagining, people will rarely if ever interact with the internet directly, it’ll instead all be through intermediary LLMs. In that world, the organisations that own authoritative information about hotel prices and locations will not make that freely available to LLMs, they will sell it. ChatGPT is trying to get ahead by encouraging them to embed themselves directly into their platform so they get first dibs on this kinda stuff before they put up the walls.

I often wonder what the internet would look like if we just banned paid advertising. Facebook, instagram, X, TikTok, they’d all have to start charging users to stay alive and I don’t think anyone would choose to pay for the brainrot. I’d like to see us remove the incentives these companies have for just gluing us to our phones.

Social media is probably somewhat responsible, but I don't think it's the biggest problem here. It's the fact that Gen Z is checking out on life, for many people there is no hope of owning a house even if you give up all luxury spending and grind, it's impossible for many on a typical job.

It's also getting increasingly more expensive to hang out with friends in physical spaces. Every business needs to pay increasingly high rents, and charge increasing amounts. You could go out to the bar and spend $100, or you could stay at home and play video games for free.

We are living in an era where the old and rich have taken over and continue to extract every last drop of wealth from the people who have the least.

“It's also getting increasingly more expensive to hang out with friends in physical spaces.”

It was a very narrow window of history, if at all where this wasn’t true. Like I spent most of my teen years at people’s houses or backyard or parks and it was fuckin great. All my best memories were spent doing nothing with people I liked. Even my clubbing years while fun were relatively forgettable compared to the mischief of running around with my teen friends not spending money.

Smoking has gotten more expensive though, maybe we should subsidize cigarettes for young men.

> Even my clubbing years while fun were relatively forgettable compared to the mischief of running around with my teen friends not spending money.

When people say that hanging out is getting expensive they didnt mean when they were 10-15 or so. Its easy to not spend money at those ages. Its not when you're 20+. You cant run around the neighbourhood anymore, or eat stuff your parents bought

For most people in my parents generation, going out to eat at all was a luxury. Some of the most tightly knit cultures of the world are also the poorest.

I was a terminally online youth in 2000s, both before and after social media and proliferation of smartphones.

Money is not and issue here. I was a middle class youth in a developing country, and internet was expensive. People who didn't have the means simply didn't go online. Contrast with the present, even lower income people have smartphone with free carrier provided Facebook. Radicalization is much easier now.

Ugh, capitalism

> We are living in an era where the old and rich have taken over and continue to extract every last drop of wealth from the people who have the least.

Exactly as capitalism always intended. We are finally reaching the dream of the system, arent we happy all!?

I feel like this is a lazy answer because there have been plenty of examples of eras where wealth equality was much better, for long periods of time. And they weren't periods of radical communism or whatever.

We just didn't tax the middle tier of workers so intensely while giving everything for free to the ultra rich. That isn't really a part of capitalism itself. It's just the specific scenario we ended up in today.

I don't think it would change at all. Look at creators on Youtube. The majority are clickbait and other crap, people trying to get money and/or influence. It doesn't require advertising by the platform. All it requires is the internet itself and nearly everyone on it. Ads might have accelerated things but the basic incentives are "more viewers = more money / influence", with or without ads. And those incentives eventually lead to where we are.

Creators on YouTube do it for monetization. There are now multiple generations of youths for which “influencer” is their desired career. If you make it, you make it big.

They might pull in more money with sponsorships but they only got there because the algorithm put them at the top where the money is.

Don’t believe me? Look up the woman who shot up google HQ when they demonetized her channel.

I know someone who got addicted to the influencer money in Tiktok and dropped out of a med school program because making 5 figures a month, taking videos of you wearing fancy clothes was way better than slogging thru med school

For those of us online in the 90s we don’t have to wonder

It was literally a utopia before business came along. Every site was built from passion, with no expectation of getting anything in return. It was a global community centered around sharing knowledge.

I’d go one step further and ban the consolidation of platforms by billionaires. The open internet no longer exists or will ever exist again

I am younger, being part of online forum communities in the late 2000s (primarily video games and art centric ones) and I truly pine for the era of the internet free from algorithmic feeds, infinite content scrolls and profit incentives. There was some profit incentive, but it was purely for administrative purposes to keep the site up.

> Every site was built from passion, with no expectation of getting anything in return. It was a global community centered around sharing knowledge.

It's a bit of false nostalgia but also it was the era of early adopters. Their concentrations in new spaces always make them better because their motivations are perpetually directed outward from where we are as a civilization. They frequently have to move on as the space they create for themselves becomes drowned out. HN has attempted to remain relatively secluded, and that has been effective up to a point.

The real dilemma there is that the early adopters who make things good need isolation while the "go with the flow" crowd needs a way to support early adopters without themselves getting in the way. Just from a basic computer science perspective, the early adopters need ways to create efficient back-pressure on later adopters so that early adopters can exist without permanently being chased and drowned out by later adopters.

The internet created something analogous to a 2D plain-world where nothing was out of reach for anyone. Without creating some 3D structure, some stratification so that people who get out of the plain can more directly communicate at longer distance, the noise is so inefficient that only those who don't value their time or those who profit off of the poor connectivity will participate.

Baking fresh cinnamon toast comes to mind. Some will accuse me of becoming distracted, but it used to be more common to speak on every open-access forum as if talking to another person in the room. We still do on different formats like IRC or less serious threads of less serious places, but it is diminished as internet culture emerged. Internet culture sometimes expects you to treat every conversation as a conversation with a forum full of angry combatants. A little bit of structure would re-humanize that culture by putting us back into our more human-sized enclaves where there is no need to gatekeep and no audience to perform to.

I'd like to think parents would be more involved if they had to buy their kids internet things but they'd probably end up swiping their creditcard on the same stupid stuff.