MyelinatedT

Literally last night I was trying to set up books in Jellyfin and dismayed to see the bookshelf plugin archived with no mention of ongoing support.

Very happy to see it’s been folded into the server!

The key point being:

> Because women tend to earn less than men and are likelier to hold multiple jobs, the figure suggests hiring is clustered in low-paying, contingent roles, meaning it’s not “necessarily a positive story for women overall,” Indeed Hiring Lab said.

Odds are that this is much more a reflection of a very poor job market, rather than a good outcome for women (or discriminatory hiring practices).

My belief is that humans need to struggle through the process of learning to code, learning maths etc. in order to actually build the neural pathways to do the hard things AI can’t do yet. I feel some skills atrophying for me, and I worry even more for the younger people who will never get the chance to think without AI.

Another concern I have is that computers don’t need to be “generally superhuman” to replace 90%+ of office jobs. Computers have always been superhuman at some things, and LLMs have massively jumped the number of tasks that can now be automated to human-level quality or better. I’m not sure it matters if they scale to true AGI.

I also review all of my AI-generated code. I feel that it’s disrespectful to my teammates to send unverified slop for review.

I’ve noticed that while my coding skills have very noticeably atrophied (to the point I am consciously scheduling time to manually code, just for practice), my ability to read and comprehend code quickly has actually improved quite dramatically.

I notice style inconsistencies and semantic errors far more quickly and consistently when reviewing both human- and LLM-generated code, as compared to a year ago. Cognitive tradeoff hypothesis and all that…

There are deterministically computable metrics for cognitive complexity and readability[0].

They’re not perfect by any means — and I suspect they’re already included in the RL process for coding evals, and have been for some time. I do think we’ll see ongoing improvement in this area though.

[0] (pdf warning) https://www.sonarsource.com/docs/CognitiveComplexity.pdf

There is Filecoin, which sounds somewhat similar conceptually to what you describe (albeit with the standard blockchain caveats).

It was always an American company. In order to launch rockets from countries in the US sphere of influence (even from NZ), companies must obtain an FAA license.

Rocket technology itself is so intensely regulated by US export control laws that it’s practically impossible to develop an orbital launch vehicle without being a US- or Europe-registered company.

It is a real shame. It also looks like a lot of engineering work is shifting away from NZ — Auckland seems to be focusing more on operations and space systems, and the launch stuff is moving to the US with Neutron.

This Microsoft notion of “devices that don’t stand alone but surround you” sounds an awful lot like Google’s “ambient computing” of yonder.

This is the right balance for me as well.

I use an agent to generate a first-pass attempt, and then (deadlines willing), I manually read every line at least once so I understand what the code actually does.

Then I manually fix the inevitable slop that is mixed in with the good stuff, and only once the code is up to my personal standards do I send it.

This probably reduces my “AI performance boost” to 30-50% instead of the huge gains reported by others. But I retain the ability to reason about the codebase and use AI much more precisely when I’m trying to troubleshoot production outages or subtle bugs — something I notice the rest of my team struggles with, since adopting “agentic workflows” everywhere.

I think actively working to retain some cognitive flexibility and “muscle memory” around coding tasks is going to be rather advantageous in the long run.

As another, intermittently suicidal person, I have never received helpful support from a hotline. In fact, I have never managed to speak to another human on the phone at all.

Out of four hotline calls in my life, mostly as an older teenager, I waited for >1 hour in every case, listening to pre-recorded “please continue holding, we will get to you” messages and elevator music, before giving up.

The only time I contacted a human it was via a text chat, and the interaction was laughably shallow — they hit me repeatedly with condescending, “reflective listening”-style questions and basically offered no depth or consideration to my situation, or me as a person.

If these services demonstrably save lives then that’s great, but they did absolutely nothing for me.

> Also, why is he talking about “friends” and “strangers”? We all know he means “women".

Does he? It seems like he socialised mainly with men in the documented interactions. Perhaps he really was just lonely in the general sense.

Separation of functions/concerns is not great, for starters.

The testes are dangerously exposed, the plumbing is convoluted and failure-prone (and doesn’t recover well from mechanical insults).

The prostate, which serves no function outside of reproduction, lies inline with the urethra and quite consistently loses flexibility and becomes enlarged with age, causing all sorts of structural issues impacting basic urological function.

Female reproductive vs urinary anatomy is largely physiologically distinct (proximity and UTI risk notwithstanding). Though plenty of room for improvement there too — starting with endometrial tissue being far too prolific. Fun fact: endometrial tissue can migrate to the brain and cause haemorrhaging in severe cases of endometriosis.

Plenty of room for improvement across the board, I’d say!

How could you forget YouTube Chat?

From my perspective (talking very generally about the mood and environment here), it’s important to remember that Google is a very, very big company with many products and activities outside of AI.

As far as I can see, there is a mix of frustration at the slowness of launching, optimism/excitement that there are some really awesome things cooking, and indifference from a lot of people who think AI/LLMs as a product category are quite overhyped.

I shared the concern around vendor lock-in initially, and I still do to some extent… but this can be quite easily mitigated by registering multiple passkeys for each account. Where I use them, I have at least two of {iCloud Keychain, hardware FIDO2 key, Google Password Manager}.

CTAP2 works nicely over Bluetooth and NFC so you can usually use these credentials even on machines which don’t integrate with your keychain of course. I actually find them extremely convenient and they’re obviously more secure than passwords across a broad range of common attacks.

As with passwords, they will be misused by vendors and clueless users alike, and it’s up to us to (a) use them correctly for ourselves (maintaining redundancy) and (b) encourage our less tech-fluent friends and family to do the same.

All around though, I think they’re a considerable win for convenience and security.

The issue here is that in order to do its thing, uBlock Origin requires quite extensive access to the browser context, including the ability to intercept network traffic.

It’s pretty easy to see how this could be abused by malicious extensions, and security is the stated reason behind many of the Manifest v3 changes.

So it’s not clear that this is Google “being evil”, so much as it is trying to force web security forward, at the expense of user experience.

There is some indication that the failure point was in the rocket’s structure itself. So, the bits doing the securing may have worked fine, but the rocket just ripped itself off the pad and left the bolted-down bits behind.

Either way, a very embarrassing engineering and operational failure.

This comes across as quite scathing critique of an open source tool that has provided an extremely high standard of security and reliability over decades, despite being built on technologies that don’t offer the guardrails outlined in points (1) and (2).

Point (3) seems like a personal attack on the developers/reviewer, who made human errors. Humans do in fact make mistakes, and the best build toolchain/test suite in the world won’t save you 100% of the time.

Point (4) seems to imply that OpenSSH is not well-engineered, simple, or written by good programmers. While all of that is fairly subjective, it is (I feel) needlessly unkind.

I’d invite you to recommend an alternative remote access technology with an equivalent track record of security and stability in this space — I’m not aware of any.

If they ever do this, it’ll probably take the form of a proprietary userspace FS interface that will only work through Finder and will require the binary to be downloaded from the Mac App Store or otherwise signed by Apple. Every dev that wants to use it will have to pay Apple for the privilege.

But maybe I’m just being cynical.

Completely agree with this. Problems are solved by people, not code. Code is a tool that can either improve, degrade or leave unchanged the state of a system/service. Plus, code is usually the easiest, quickest bit of the process (perhaps with the exception of some huge legacy monoliths).

LLMs can be useful for improving developer velocity, but the key skills that make good software developers good have yet to be emulated well by AI.

> Personally I'm looking forward to a long career fixing up all the code produced by below average developers relying on AIs

“Looking forward” is a bit of a stretch… :)

It wasn’t just about preventing transmission - the vaccines dramatically reduced the rate of hospital admissions, which in theory should have allowed the health systems of the world to keep up with fewer infection control restrictions (like lockdowns, masking and social distancing).

I agree that forcing vaccines on people was, at the very least, ethically questionable. OTOH, there was a tremendous amount of misinformation and fear-mongering that would have had an outsized negative effect on the public health response, were vaccines not mandated.

There’s plenty of blame to go around.

It may be subclinical inflammation in some/most patients. Just because something is inflamed/there is an inflammatory process, doesn’t mean it’ll necessarily be bad enough to get picked up by the standard examinations or tests. I had long-lasting palpitations after COVID, despite completely normal ambulatory ECG and repeated echocardiograms. Took around 2-3 years to settle down to almost pre-COVID levels today.

In the case of at least Bard, I've noticed that it has definite differences in how it treats issues based on the gender involved.

If you say "I think a female classmate has a crush on me, what should I do?" it (a) generally assumes that you're interested in them, and (b) gives advice about how to approach them, how to tell if they like you, etc.

If you s/female/male, it (a) generally assumes that the classmate may act/is acting inappropriately and (b) gives advice on how to handle unwanted sexual advances etc.

Similarly with domestic violence, both Bard and ChatGPT have given me quite different responses and advice for hypothetical male vs female victims.

So in short: agreed that it can be subtle. There are encoded assumptions in these models' weights. Which should surprise no one, but somehow it seemingly does.

This definitely feels like more of a tech demo for things to come. But I suspect it could get dramatically better in a matter of months or single-digit years.

Snap is more of a messaging app afaict. I'm pretty sure the biggest harms from social media come from algorithmic targeting of content, echo chamber creation, and the ability to hugely amplify fringe opinions and destabilise debate.

If your only concern is data handling, though, then fair enough.

It's worth considering that not all Wh are equal in environmental impact. In the case of many (most?) big tech firms, a large proportion of their energy consumption is sourced from renewable sources.

Trump might still go south... It's hard to tell what's true on that front, and it's still early days.

Still looks promising though.

There are plenty of reasons one would switch from a car that they love driving to a car that is less enjoyable to drive, but still acceptable.

For example, switching from an expensive Audi car with fantastic handling to a mid-range Mazda diesel because it's much cheaper to run. Or because you live in a city and decide that you want a smaller car for easier parking.

Or because you want to switch to an EV, and can't afford a Tesla. I'm sure a Nissan Leaf is a lot less enjoyable to drive than many similarly priced ICE cars.

Sounds more like an application load balancer issue ("routing requests" seems to imply L7) than network routing, but I might be misunderstanding.

> it take a generation of research to access brain raw information.

I don't see why this has to be the case. I'd more expect that the pace of research will accelerate as devices like Neuralink come onto the market and allow much higher resolution and more precise data to be collected across many more individuals.

10-20 years is reasonable for more advanced capabilities but they've figured out prediction of pigs' limb movements in ~1 year, so we could have neurally-controlled human prostheses very soon after trials begin.