willidiots

Just using LLMs enough I've developed a sense for the flavor of writing. Surely it could be hidden with enough work, but most of the time it's pretty blatant.

They do acknowledge that in the notes, referencing one other study (n=3) with one healthy aged author that did not exhibit these signs.

Relatedly, as this is a single-case study, without a control author, it remains possible that the decline in lexical diversity reflects natural ageing. But note that previous research [15] has found that healthy authors can maintain stable linguistic diversity into their late 80s, suggesting that the decline observed in Pratchett’s TTR is indicative of pathology rather than typical ageing.

IMO still all very inconclusive but an interesting avenue to explore.

Quoting TFA: "It’s worth noting that none of this was really legal; the law technically stated that TikTok shouldn’t have been allowed to exist for much of this year. Everyone just looked the other way while Trump and his cronies repeatedly ignored deadlines and hammered away at the transfer."

Google TV showing ads on the home screen convinced me to buy an Apple TV. Had to go back to a set-top box - to use the same apps I have built into the TV - just because Apple won't spam me with this shit.

I have them and like them. I don't wear them constantly, but on days when I'm doing something interesting, they help me document much more than I otherwise would.

To TFA's point - "Bars" are relative and relatively meaningless - [SS]RSRP, RSRQ and SINR are your real numeric signal strength / quality measurements.

Not sure about Apple, but on Android, individual carriers can set the number-to-bars thresholds. Two otherwise-identical signals could be represented as a different number of bars depending on your particular carrier: https://source.android.com/docs/core/connect/signal-strength

I made a conscious effort to switch from hyphens to em dashes in the 2010's and now find myself undoing that effort because of these things, so I try not to instantly assume "AI". But look long enough and you do notice a "sameness": excellent grammar, fondness for bulleted lists, telltale phrases like "That's not ___, it's ___."

And a certain vacuousness. TFA is over 16000 words and I'm not really sure there's a single core point.

I'm a big fan of the MWCBTY keyboard format, it's especially efficient when you have to type a lot of G's.

Snark aside, I think it's laziness and the shotgun approach. The author writes some rough thoughts down, has an AI "polish" them and generate an image, and posts an article. Shares it on HN. Do it enough, especially on a slow Sunday morning, and you'll get some engagement despite the detractors like us in the comments. Eventually you've got some readers.

The whole thing feels AI-padded.

It's not black and white. My SO doomscrolls facebook on her laptop for hours daily. Certain parts of creative workflows are better on phones (or other devices) than laptops - the article acknowledges this in the "hybrid workflows" section.

IMO the important thing to be mindful of is your creation-vs-consumption balance. We tend to overindex on consumption.

I had one of the second-gen units, it was programmable locally, and you could locally enable the learning mode (which was not good)

I agree that libraries are a thing, all problems in CS can be solved with a layer of indirection, etc. I also have no issue with AI-gen projects if they're good.

In this case, they posted a README full of nonsense diagrams, didn't fix the broken characters in their UX, and breezed over the complexity of the dependencies (ESP-CSI is very cool but requires specific hardware, with two ESP devices and external antennas). Feels sloppy.

The interface and code smacks of Claude. It's basically someone's AI pet project wrapping legitimate third-party tools.

The Cascadia Subduction Zone earthquake of 1700 was dated using Japanese tsunami records: https://ui.adsabs.harvard.edu/abs/1996Natur.379..246S/abstra...

Meanwhile, violence remains perfectly acceptable.

Low-latency LLM for my home automation. Anecdotally, Gemini was much quicker than OpenAI in responding to simple commands.

In general, when I need "cheap and fast" I choose Gemini.

Some of them literally have "visualelectric" in the filename.

(ex: https://images.squarespace-cdn.com/content/v1/67160715a896dc...)

Yes! There are a whole series of posts with photos and maps. Click the red "Tumangan, we are coming!!!" link at the bottom of the first post to jump to the next one, etc.

Half of the battle is avoiding over-documentation - I only write things here that I know I'll want to remember later: for me these are important notes, figures I hear/see in meetings, critical names of people, that kind of stuff.

Mostly it's just bullets. If a bullet is critical I'll prepend it with !!!. If it's a task, I'll use * instead of - for the bullet marker (then move the task to my separate task tracker later - notes doc isn't good for long-term task management)

If I want to tag the note explicitly with a "tag", I'll add one in brackets - but honestly these aren't that helpful.

Ctrl-F will generally find what I need. Aside from that, what's mostly been helpful is the dates - I often remember roughly when I learned a bit of information, so if I can't find it with Ctrl-F, I'll scroll back to that "era" and look around.

Pay a lot of attention to yourself when you're looking for historical information: what do you search for? Are there key things you often look up over and over? If so, what search terms did you use when trying to find them? Add those explicitly to the entry, so next time you'll find it easier.

I use a flat text file called "notes" on my desktop, and I leave it open in Sublime Text in a corner of my second screen. Periodically I throw a datestamp in there as a reference point. For generic "stuff that should be retained manually" it works well - easy to add to, easily searchable.

One thing to be mindful of is that this limits your ability to help your partner as you age. State intervention can play both ways.

I literally had a 90 year old woman approach me at the dentist yesterday, complementing my Model 3 and telling me "I just love Elon. I had a dream about him last night!" For these types, it's a cult of personality, not logic or policy.

I was also at the gun range last week and overheard a conversation between two Trump supporters. They were outraged by his behavior since taking office, and said outright "if we had to vote again right now, half of us wouldn't vote for him".

We've had good luck with Mosquito Dunks, both in natural standing water sources and in dedicated "trap buckets" filled with water and vegetation, but you have to do it early in the season as the dunks only affect the larvae.

Last year I came across a water-filled stump that was teeming with larvae swimming around in it. Dropped a piece of a dunk in, came back 24 hours later and the water was clear.

https://www.youtube.com/watch?v=TqZpi8zAqe0

This little video is such a great tribute to him. RIP David.

It's definitely a real, known threat. DARPA and friends have been working on their own swarm technology for years: https://www.youtube.com/watch?v=W34NPbGkLGI

There already was! It was called Passpoint R2 Online Sign-Up, but it never got traction on the phone side of things, so it's now being deprecated by the Wi-Fi alliance.

It's really a business problem. IMO you shouldn't have to solve this just because you've gone indoors – you already pay a carrier for connectivity – but many carriers don't want to own that responsibility.

Looking at the charts, the Cessna brushed right up against the ground-4500 class C space. Not the smartest place to be off-frequency, in what sounded like mixed viz conditions. At the same time yeah, Austin Approach screwed up - "that traffic's through your final", when obviously it wasn't.

TCAS will use the Mode C or Mode S altitude provided by a standard SSR transponder if the intruder aircraft isn't TCAS-equipped. To determine the appropriate maneuver, the system calculates which trajectory will increase the vertical separation at the closest point of approach. TCAS generally favors minimizing disruption by requiring the smallest possible deviation from the current vertical trajectory, but if the intruder is climbing or descending, the system biases the RA based on the threat's current altitude trend, with the expectation that the intruder might level off.

You can read all about the threat resolution logic in section 3.10.8 of ICAO 9863.

Grew up rural, but spent a lot of my formative years on BBSes and this budding "Internet" thing.

Moved to SF, spent 20 years there, then moved back to the countryside because I was going to climb mountains instead of tech and breathe the fresh country air. Hiking and climbing was a great adventure but 5 years later I see that it wasn't the "solution" I was expecting it to be.

Sometimes I feel the way you're feeling; I miss the tech scene, the vibrance of the city. Other times, I feel great about the move. I know more people in this tiny town than I ever knew in the bay, and I feel more invested in the community's future. I've grown a lot through this change, and learned much that I'd never have learned in the bay.

I do wish we'd moved closer to my parents. Caring for them as they age is difficult from afar.

The grass is always greener, somewhat, and IMO the best "answer" is to make what you can from wherever you are. Don't be in a rush to force life's progress.

How about the rest of the eastern US? Billion dollar climate disasters are hitting everything, not just coastal cities in Florida and WUI towns in California. There was just massive flooding in Tennessee due to Hurricane Helene - are they at fault for living in "hurricane country"?

Climate related disasters are intensifying everywhere. What's now "inevitable" may have been unthinkable when some of these folks moved there.