Thank you. But he was two months shy of his 90th birthday and, except for struggles with osteoarthritis and Parkinson's, he had a good run. I'm sad that he's gone, but it's not like it took anyone by surprise. And I'm glad that his suffering, which towards the end was not insignificant (though he was very stoic about it), is over.
The last week has actually been pretty (ahem) interesting in a lot of ways. I should probably write a blog post about it.
I suspect people will get pretty riled up in the comments. This is fine folks. More people will make their stuff machine-accessible and that's a good thing even if MCP won't last or if it's like VHS -- yes Betamax was better, but VHS pushed home video.
That's what I don't get with AI, isn't it supposed to make us work less? Why do I need to bother making my websites AI friendly now? I thought that was the point of AI, to take something that's already there and extract valuable information.
Same with coding. Now I don't get to write code but I get to review code written by AI. So much fun...
AI is not great at browser use at the moment and it's also quite inelegant to force it to. It's one thing if it reads your nicely marked down blog, it's another for it to do my groceries order by clicking around a clunky site and repeatedly taking screenshots. Not to mention how many tokens are burnt up with what could be a simple REST call.
So to answer your first question, it's less about _reading_ and more about _doing_. The interfaces for humans are not always the best interfaces for machines and vice versa in the doing, because we're no longer dealing with text but dynamic UIs. So we can cut out the middle man.
As for coding, Karpathy said it best: there will be a split between those who love to code and those who love to build. I too enjoyed writing code as a craft, and I'll miss doing it for a living and the recognition for being really fast at it, but I can do so much more than I could before now, genuinely. We'll just have to lean more into our joy of building and hand-code on the side. People still painted even after the camera was invented.
I’m all for making data more machine accessible, but it’s not like there was a shortage of ways to implement that. Hell, if most sites implemented OpenAPI, there’d be no problem to solve.
The choice of whether to make one’s service open to mechanical use is a business decision. Imagine a world in which YouTube could easily be accessed by scripts. Google does not want this; they want quite the opposite.
Yes, when I said Betamax I was actually referring to Swagger/OpenAPI. It's been around for a while but it didn't catch on the way MCP did.
What I'm saying is that the AI hype is making people make that business decision, and that is ultimately a good thing because it means more human accessibility. Not just for people with disabilities, but through interoperability and fewer silos like YouTube.
We’re talking about agents here. (These are, after all, what MCP servers are meant to serve to.) Thus we’re talking about the need for services to be efficiently agent (computer) accessible, not efficiently end-user accessible.
I came to a similar realisation about world news a few years ago and live a much less stressful life now. Especially since most of the news was about the US, and I don't even live there and there's nothing I can do about it. If something really important happens, eventually I find out from friends or family.
Same when it comes to staying on top of tech news -- almost everything is a flash in the pan. I used to bookmark cool new products, never revisit them, and then a year later realise half of the links are now dead.
One thing I realised though is I still need to mindlessly browse an endless feed every once in a while for some downtime. One way or another I'll want to fill that time with something, so it's a question of being mindful what goes in it. So my drugs of choice are Hacker News, and carefully curated YouTube subscriptions.
I currently use Hapi (https://github.com/tiann/hapi/) for this and find it quite handy. I can easily tap into a session on my PC from my phone.
Before that I used Happy (https://happy.engineering/) which is also open source and a lot more sophisticated. It has a voice assistant that can chat with Claude Code on your behalf in the mobile app. However, it wasn't very reliable, and there are other reasons to use Hapi instead (documented in the Hapi repo).
Before that, Omnara (https://www.omnara.com/) a YC company and seemingly a proprietary Happy fork (?) but it never worked properly for me.
Long story short, there are a few of the around, and frankly I really like to use them. Unlike other commenters, I don't find that they wreck my work-life balance. Rather, I can go out and have a walk in the park, only checking in on long-running tasks every once in a while. The diff view is pretty good too. There are many tasks where I'd rather not stare at my PC all day and instead do other things, and these tools allow me to do that.
Also recommend Opencode which has `opencode web` built in. It's really impressive how good opencode web is. It's far more polished than I'd have expected from a free alternative!
What.cd were extreme sticklers for quality! When you applied to get in, they did a live interview on IRC to test your knowledge of ripping, transcoding, and different kinds of compression, how torrents and private trackers work, and their code of conduct. I remember studying for it. They also had ways to make sure you weren't cheating like checking your screen, as well as very aggressive automated checks for VPNs and blocklisted IPs to prevent ban evasion and multiple accounts.
They also had good incentive structures for keeping the bar high -- you could get kicked out for having a bad ratio, so the easiest way to pump your upload up was to fulfil obscure requests for FLACs you could purchase online but were extremely difficult to purchase (if you're lucky it's just an unknown artist on Bandcamp). I discovered a lot of obscure music this way, some that I'm still looking for to this day after it shut down.
Because I cared so much about being part of that private tracker, this is what also prompted me to rent a seedbox for the first time. I paid in Bitcoin out of paranoia (I lived in Germany where the fines for piracy are HEFTY, and they actually do come after you) back when Bitcoin wasn't really worth that much, and later found that that old wallet suddenly had a couple thousand in it instead of the spare change I couldn't move!
I must have joined at a different time because all I needed for me, a total annoying script kiddie leech, just needed an invite code (or link? I forget)
The author may like https://omnara.com/ (no affiliation) instead of SSH-ing from their phone. I have a similar setup with Obsidian and a permanently-on headless Claude Code for my PKM that I can access through the phone app.
I haven't dug into all the sources, but I think there's a potential confounder here, or maybe even reverse causality. The author seems to assume causation when the studies only indicate correlation. E.g. the first link says "chronic loneliness increases mortality risk" but the actual source says "actual and perceived social isolation are both associated with increased risk for early mortality".
So for example, it's possible that if you already have chronic illness, a disability, or any other kind of health issues, you're more likely to have higher social isolation and therefore be more lonely, in addition to having a higher mortality risk. There's an outside variable (your health) that is correlated with both (loneliness and mortality), but that doesn't necessarily mean that loneliness causes mortality. If this were the case, we could defend claims like "autism increases mortality", because we already know that autism increases social isolation.
Poor health increasing social isolation isn't even a hard casual path to argue. Common health problems can physically restrict how often and how long one leaves the house: people on oxygen can only travel as far as their supply and weakened lungs can take them, or people with bowel disorders might be reluctant to be do anything without easy and discrete bathroom use, people with visible symptoms might be embarrassed and avoid socializing.
The loneliness-associated protein study linked in TFA doesn't seem to control for health status. So preexisting conditions may have affected the correlations.
Yes, this seems blindingly obvious to me. Maybe because I have chronic health problems myself. Perhaps those blessed with good health are blissfully ignorant.
When you are chronically ill, socializing falls pretty rapidly down your list of priorities.
That said, it seems pretty clear that incentivizing healthy socialization can significantly improve quality of life, reducing perceived suffering and quite possibly improving outcomes through reinforcement of the various social needs we have.
We are fundamentally social primates; there are prior studies demonstrating the overt harms of isolation and ostracization in humans, primates, and other mammals. Our immune systems, metabolism, and cardiovascular health are all tied intimately to things like stress and hormones, with feedback effects that can amplify disorder, or suppress it, depending on the directionality of the inputs.
I see the "just so" element of explanation to this study, but I think that even if the underlying causal factors are more complex, it is so directionally correct that I have no problem with the conclusion of the study, even if it's not correctly justified scientifically.
And don't forget that physical loneliness, that is, actually being alone, eliminates one major feedback source that something could be wrong with your health, or a source of immediate aid if e.g. you go into cardiac arrest.
Maybe the researcher above touches on these things, but more generally, there should be a standardized probability and statistics exam for ALL aspiring scientific researchers, and a high score should be the minimum cutoff. The influence that a statistically flawed study can have over our collective futures is too dangerous.
> The influence that a statistically flawed study can have over our collective futures is too dangerous.
An even bigger danger: with all of the flawed / p-hacked / over-hyped studies, the public (and the legislature) will start to believe that NO science is real.
It worries me how much argument there is over things I consider to be facts. And how much effort goes into undermining science when it is not in the corporate interest (eg cigarette manufacturers funding “inconclusive” studies).
The first source I clicked into was a meta-analysis of randomized clinical trials.
What you say sounds true about chronic illness and isolation. These researchers are looking at research done using actual interventions and real results.
What should they do to analyze this more than RCTs and then meta-analysis of RCTs?
Another huge factor is probably lack of feedback loop. Its surprisingly easy to press on by yourself through various health issues until its too late. However someone around you might say you look pale recently you should see doctor as an obviously contrived example.
My dismissive but practical take is "well yeah there's nobody in the room to call an ambulance when you have the heart attack you'll most likely have", which mindfulness classes and support groups don't help with. There's practical benefits to having people around.
There was a funny statistical artifact I read once (well, for some value of 'funny') that home was a terrible place to have a heart attack, because people are willing to 'just lie down and see how I feel' rather than in a restaurant or movie theater where an ambulance will be called.
It would be interesting then for similar studies to add a dimension in addition to self-reported loneliness on self-reported time spent outside the home. While it's likely that sedentary elders are lonelier, it's not a 100% overlap (some elders live with close family and some lonely people are still very physically active/outside). You would expect lonely, active people to have lower death rates than lonely, sedentary people with similar pre-existing conditions under your hypothesis, and it would be a powerful thing to prove because it's a lot easier to make people go out than to make them make friends.
Or related: there's no one to nag you about going to get that funny ache checked out. Men particularly are notoriously reluctant to go to a doctor for various reasons but a worried partner might persuade them.
By these statistics there aren’t enough healthy people to provide care for those that are less healthy
The challenge here is that healthy people don’t desire to be around unhealthy people.
Society provides no incentive or social benefit for otherwise healthy people to be around the unwell to call the ambulances. Even as a nurse, hospice worker or caregiver, the pay/benefits are non existent for the amount of emotional and physical labor needed for care.
You don't have to be that healthy to call an ambulance! As long as not everyone in the sick-and-poor commune has a heart attack at the same time this should be a safe enough system.
Ambulances are expensive enough that people are hesitant to call them, sometimes even in life threatening situations.
And if the person is unsure whether the situation is critical, they might try to "sleep it off" rather than driving or getting a ride, because ER is also kind of expensive and you could be stuck there all day.
I once did a course with a paramedic on basic aid. We were discussing choking, which is a condition that really needs a 2nd party to intervene. Someone asked what to do if you live alone (with no close neighbours) - the answer was essentially ‘good luck’
The idea of the heimlich is to put sudden force on the diaphragm and force air upward. You can do that alone by pushing your upper abdomen against a chair back, counter, railing, whatever. Not something I've ever tried, but good to know about in case.
I know how to in theory, and I think I'm probably "above average" at calm-in-crisis, but my confidence that I'd calmly rescue myself via self-heimlich while unable to breathe is not high.
>Recent meta-analyses examining 2.2 million individuals across 90 cohort studies reveal that social isolation and loneliness trigger measurable biological cascades comparable to traditional disease risk factors.
Its frustrating, because cohort study experimental designs like these can in principle chip away at reverse causality (i.e. observe loneliness exposure before a cardiovascular disease prognosis, compare difference-in-difference between treatment/control), but the meta-analysis doesnt clearly state whether this constraint was applied. But even a study like this would have issues with medical participation, so that would need to be controlled, preferably with a prospective design.
Having a spouse or friend nag you into going to the doctor is undoubtably a part of the effect, both for the practical side and for the feeling that someone would care if you were ill. (Assuming this is what you mean by "issues with medical participation.")
As an anecdote, I have a couple of elderly family members who did not mellow with age. Instead they became even more toxic and abusive as they approached their 80s. (I have another who was the complete opposite.). The abusive ones hurled curse words as soon as you entered the door; then they "switched" personalities and acted as if nothing happened. I also know of an acquaintance whose son had the similar personality traits even though he's only in his 40s. The son needed a 24 hour a day caregiver because he's mentally challenged. In short there are a whole host of physical and/or mental problems that confound the situation.
>The author seems to assume causation when the studies only indicate correlation.
once I heard Feynmann say in a youtube video that (paraphrasal) "we don't know what causes gravity, we just know that it exists, it's a property of matter"
then I realized, our experiments never show causation, they only show correlation. gravity has 100% (in our experience) correlation to matter. admittedly, that's a pretty good correlation, but for all we know, gravity causes matter. energy too, apparently.
Philosophers have worried for a long time about whether we can actually observe causation.
David Hume was famous for arguing in An Enquiry Concerning Human Understanding that we can't observe it and we instead have a "custom" or habit of expecting effects to follow causes.
> After the constant conjunction of two objects—heat and flame, for instance, weight and solidity—we are determined by custom to expect the one from the appearance of the other.
Religious philosophers have sometimes gone to the extreme of occasionalism, where they've maintained that patterns and regularities in nature were just habits or customs that God chooses to follow:
I was not aware of this, but this actually supports my point, since the reasons in that study seem to be correlated physical comorbidities, so that makes the claim "autism increases mortality due social isolation" both easier to make (as you can misinterpret the stats) as well as less defensible.
Social isolation is also not the same thing as loneliness. The paper recognizes this and gives different outcomes for three distinct categories (social isolation, loneliness and living alone). The blog post mistakenly cites the outcome for living alone as the outcome for loneliness.
If you think things through like that then you can only ever attribute physical, material, biological factors as the only things involved in mapping the causation. In other words, every time we attempt to draw the line to mortality, it gets hacked down as we keep reducing the argument.
We cannot explore the possibilities of truth if we do that, but I can appreciate the due diligence. It’s a tricky subject, but life experience informs many of us that there is something more going on than “I materially feel like shit”.
There is a taboo element to loneliness that isn’t often discussed, and that is “I feel hurt that I can be left alone, or that anyone can be left alone or isolated”. So, while the source of the isolation could be material, the feeling that manifests from it is an actual hurt that one feels from the actual thing (isolation). For example, we may be killing our elders when we isolate them in care facilities.
I can’t say if we have the sense as a society to accept data that suggests this pain can be linked to mortality. Isolation in itself isn’t the killer, it’s the pain of “well how could any society leave anyone alone”, and such a phenomenon can be witnessed in the macro outside of yourself (how can we leave people on the street? Etc).
Loneliness and isolation is often in sequence, after abandonment, or negligence, or unforgivingness (if the person “deserved” the isolating). A phantom, immeasurable pain. And even more painful, to deny it afterwards.
There is considerable evidence that people who live alone are at greater risk of hospital re-admission, probably because they are less able to care for themselves properly.
Of course there is no direct causation between loneliness and death. You don't directly die from lack of social interaction, but you do directly die from lack of food. However, there is a clear causal link between loneliness and habits that increase mortality.
> Of course there is no direct causation between loneliness and death.
Of course there is. If you are alone at home, who calls an ambulance if you have a heart attack or similar condition?
If you are living together with someone, the chance they are arund while it happens is all thats needed to skew the statistics.
The APIs thing: it's not just Chrome, but Chromium too. I first noticed this when trying to replicate some of the screen sharing UI (buttons to share different things) from Google Meet, only to find that no non-Google domains have access to those APIs in the Chromium source code. Made a huge deal about it but nobody seemed to care.
reply