Instead, TFA begins addressing objections and potential misinterpretations before even making it's case.

There's something even worse, taking my 97 year old mother to her gerontologist :-). Her HMO has a group office of 8 gerontologists with a shared waiting area which easily seats >60, but is always packed since most 80+ year-olds treat doctor appts like plane flights and arrive an hour early. Most of the patients can't hear well and take calls on speaker phone at FULL VOLUME, then constantly remind the caller to talk louder.

It's maddening enough when I go maybe once a quarter and sit there for an hour but imagine working in that room every day. Last year the reception desk relocated closer to the elevator and a very solid wall with thick auto-close door was constructed in-between. Perhaps OSHA has a 'cruel and unusual' designation :-)

[Empathetic side note: hearing loss in the elderly is a serious hardship which not only socially isolates but also accelerates cognitive decline while decreasing independence. According to several audiologists we've consulted, my mom isn't an outlier. It's normal that even state-of-the-art, high-end (>$5,000) hearing aids, properly calibrated, only help a limited amount at >80+ yrs old. My mom, despite occasionally struggling to remember an infrequent word, is still mentally sharp. Yet it's rare she can understand people on the phone well enough to make her own appts etc. I'm really hoping DSP, MEMS and AI offer better hearing solutions before I need them.

Inevitable hearing loss combined with forced 'app-ification' and mobile UX trends are increasingly making the modern world inaccessible to the very elderly, cutting off even those who are mentally capable. Talking to the staff at my Mom's independent living facility, it's a shit show and getting worse with every major OS update. They've been forced to staff a Mobile Phone Support Desk. I really wish Apple & Android UX teams were required to work a week at that desk.]

To be clear, I have no problem with math people conveying math stuff with math notation but it seemed like the type of article which, even if it will get into notation, first at least explains "the main idea of it" without notation, then expands with notation as necessary when fleshing details or implementation.

There are some McLaren street cars that use passive venturi channels which from their shape alone, generate more down force at high speeds than the car's weight. In theory, the car could drive upside down (if there was a Daytona-sized race track shaped like a ~30 ft diameter pipe).

However, those assuming it will never be sold or even legal to be sold may not be aware there already exist astronomically expensive, completely impractical, quite unsafe, track-only hypercars which are not legal to drive on any street.

As for the technical legality of the 'flying' features ask your LLM about: the Ground-Effect & "Sustained Flight" Jurisdictional Boundary in FAA / NHTSA regs, Off-Road Track-Only exemptions from Federal Motor Vehicle Safety Standards, and PHMSA / DOT-SP (Special Permit) authorization for consumer COPV pressure vessels. Making the "hopping" version for sale is a wildly impractical marketing stunt but they can probably do it, although it'll cost far more and ship much later than whatever they promise and will sell at most a few dozen.

I've found discussions on HN tend to quickly get pulled toward polar extremes between those who are justifiably impressed at the increasing effectiveness of LLMs and those who highlight the brittle jaggedness and counter-intuitive gaps of LLM 'intelligence' in daily practice. One side keeps citing more examples of 'uniquely human traits' LLMs don't seem close to replicating while the other complains of eternally moving goalposts.

Today I think both extremes are partly right but also partly wrong, and end up largely talking past each other because so much remains ill-defined, uncertain or currently unknowable. IMHO, the most accurate position is a carefully nuanced middle-ground built on well-defined scopes, terms, time frames and probabilities with frustratingly wide error bars. Partisans at either extreme seem uninterested in more than drive-by hot takes.

It can be simultaneously true that "Someday AI will be able to functionally replace all human knowledge work" and "That day does not appear to be soon, and even when it arrives, there will still be uniquely human traits which are valuable and that LLMs can't satisfactorily replicate."

I've read Kahnemann's book and how I've seen System 1 / 2 typically used in the context LLMs often makes them seem like poor fits to what's being referenced. The concept was created to broadly group types of human reasoning but it never struck me as rigorously defined enough to be a technically precise distinction. Applying what was a mildly useful directional thumbnail referencing human thinking styles to LLMs, which can mimic human output but via completely different processes, quickly goes off the rails. As we get more detailed and precise, the System 1 / 2 distinction seems to increasingly diverge from how LLMs behave.

A quick example is how basic Flash-type chat models without 'extended thinking' or MoEs will keep suggesting different solutions instead of stopping after a few failures to step back, gather more context and diagnostic data and evaluate changing approaches or questioning the question. When I do that, the 'interrupt' is often a 'System 1-ish' intuitive feeling like "I suspect I'm barking up the wrong tree" then, once stopped, I'll switch to System 2 to analyze. But in LLMs that "interrupt" behavior typically arises from adding more System 2-ish thinking. N=1 but that's not how I usually 'get there' and is orthogonal to what the System 1 / 2 model suggests.

I really enjoy your writing on StoneTools. Sorry to post 'housekeeping' here but I couldn't find a way to reach you and it may be helpful in solving why so many people here on HN and over at TwoStopBits have reported trouble reaching your site due to bogus security errors.

Over time, I'd just accepted that I'd never be able to reach your site again but I was really interested to learn more about Lotus Improv on the Next! So, today I spent an hour with Claude.ai doing command line testing through various internet connections using different ISPs and determined that in my case, it's because my ASUS router, a quite popular brand used by millions, comes with a security setting turned on by default, which uses block lists supplied by Trend Micro and "Stonetools.ghost.io/" is blocked under the category "Phishing."

I'm sure this is a false positive, as there's nothing of concern on your site but it still probably impacts quite a lot of people who get a weird error and have no way to let you know. Unfortunately, there's no way in the browser to get around this error nor any way on an ASUS router to manually whitelist a site. I did submit a request to review your site at Trend Micro's website but it's much stronger if you as the site owner do so. I also asked the LLM to prepare the summary below. I hope it helps and please keep up the great work. We need more good retro computer writing!

====================

What's happening: a company called Trend Micro runs a popular website-rating service which many home routers and security products include by default. Trend Micro currently flags stonetools.ghost.io/ as "Dangerous" saying it may contain malicious software or phishing. Any router using that service interrupts the connection and, unfortunately, the error which appears in a user's browser contains nothing indicating it was blocked at the router, only a cryptic message saying "Secure Connection Failed" and "SSL ERROR RX RECORD TOO LONG", making it look like an issue with your site. Disabling the blocking requires changing four different router settings on different screens. The big "AiProtection On/Off" button alone doesn't do it. That's one reason it took a while to identify the root cause.

How I confirmed it: After tracking down all the security settings and rebooting several times, your site now loads fine! It also loads fine from a phone on mobile data or through a VPN, so your site and your host are working. You can see the rating yourself at https://global.sitesafety.trendmicro.com/ and enter stonetools.ghost.io.

What to do:

1. Check your blog for anything odd: unfamiliar links, scripts or code injections you didn't add, and embedded content from other sites (old posts too). If you find anything, remove it first.

2. If the site is clean, go to that Site Safety Center page, click Check Now, then Reclassify Request, choose the rating Safe, and submit. Trend Micro will email you a confirmation link, so click it. Then wait for their reply, which can take a few days.

3. Forward this to Ghost support too. They host many blogs and may be able to get this reviewed faster, and they can tell you if anything on their side might have triggered it.

Until it's fixed, readers who can't get through can try their phone on mobile data, or turn off web filtering in their router.

As a serial tech entrepreneur with a few decades of hands-on experience both in my own startups and also spending some years as a VP / product group owner inside an F500 acquirer, you may want to think more deeply on this. A hidden downside of modern A/B testing analytics is how easy they make it to min/max short-term gains while masking delayed and second-order effects.

In this instance, no doubt the analytics show this bumps ASP in the short-term but those analytics can't show you the downstream impacts. These tend to accrue over time creating an overall drag on the business that you have no way to identify. A good case in point is how Adobe successfully min/maxxed their way to a large portion of their customers either not trusting (or hating) them. The brand damage is now likely irreparable. Do what you must to make payroll but you're on a slippery slope. Be careful you don't let 'any dark pattern is okay, as long as it's legal and works' become status quo. The consequences can accrue faster than you can exit.

OAI and Ant got blowback for essentially strip mining mathematical discoveries into PR wins for their models, accidentally laying waste to the research field in process. Throwing ~$5M each into PR whitewashing (partially in token credits), seems self-serving.

Kudos to the Simons Foundation for seizing the moment to offer the labs a fig leaf of respectability. Might have been the fastest "Yes" on a grant proposal ever.

Second, the vast majority of Adobe's revenue comes from enterprise licensing agreements not individual consumers who might switch to free, open source alternatives.

Third, good, free alternatives to Adobe tools have already existed for many years, such as PhotoPea.

Well, hitting the first home run took about a decade full of base hits and more than a few strike-outs on the way to figuring out what I needed to know to do a tech startup well.

> What is the "weird" kid doing today?

Over the last decade I've spent increasing pro bono time trying to 'pay it forward' talking to classes and groups of erstwhile entrepreneurs, so I've heard versions of that question a LOT. It actually took me a couple years before I realized there is no 'right' answer because the right thing for each person differs depending their unique personality, skills, interests, context and life journey.

The most helpful thing I can share isn't a guess about which target you should aim for but rather some advice about how to approach your journey. First, I hope you want to be a tech entrepreneur because that's the only thing I know about. I have no idea what skills will be in demand to get a high-paying, secure job or how to get promoted. I had to become an entrepreneur because I was terrible at being an employee.

I never aimed for my guess at what would be "hot" next and I doubt my guess would be right anyway. Every time I found success it was by starting with a real problem some customer actually had. Initially that customer was me or a friend but it doesn't matter. It just has to be a serious enough problem that motivated customers are actively looking for new and better solutions than what exists. Then figure out a way to solve their problem better enough that they'll actually buy it. It doesn't have to be perfect in every way, just better enough in most ways and perfect in at least one way.

In fact, some of my best products became hits by cleverly solving one part of a problem SUPER well, but only did so at the cost of some odd downsides in other areas. Here's a generic example of the pattern that worked: my friends and I all desperately wanted a Zig-Zag Widget but good ones were stupid expensive and the ones ordinary schmoes like us could afford Zigged and Zagged like shit. But while we were bitching about this sorry-ass state of affairs over pizza and beer one night we theory-crafted how it might be possible to make a Zig-Only Widget ordinary mortals could afford. The dumb idea was by not Zagging at all, it just might be able to Zig even better than the best. Since we all had no Zig or Zag, that seemed like a trade-off we might be pretty okay with. Turns out other people were too... especially after we worked on adding some 'pseudo-Zag' features that worked around the worst parts of not Zagging. And that's how we came up with the world's best Zig (but no Zag) and sold it at 1/4th the price of even a half-assed Zig-Zag. Not really new technology, just a very clever mixture of the exact right old tech with a little new stuff, some clever UX innovations and cool workflow integrations. Blam! First Product of the Year.

If prototyping dozens of different iterations and trying each one out with real, live problem-having customers and listening empathetically for hours to their constant complaints, ridiculous demands and occasional awesome suggestions sounds like fun, then you may have found your calling. If you just want to write great code for a good paycheck and be left in peace, then, my friend, run the other way. :-) It's ironic because I got into making products because I loved programming but after the first few years, I wasn't writing much code anymore. It was invaluable that I'd learned the business by starting as a self-taught, down-to-the-metal, assembly language, real-time graphics programmer because I was able to identify and hire programmers far more skilled than I'd ever been but after the first million in revenue and 20th employee I personally spent more time writing or reviewing ad copy and legal contracts than code. But that's also invaluable because as the company grew and various fires flared up in every part of our business, I got pretty damn good at doing marketing, sales, PR, finance and even legal stuff myself when necessary. And all that was the best prep ever for being able to run the place when we hit a hundred million in sales and hundreds of employees.

I used to hate the constant looming threat of existential doom always hovering over our tiny startup. Then one day I finally understood that the constant stream of scary new challenges were there to force me to keep growing and learning new skills. not only survive, but to level up and feel the satisfaction of summitting ever higher peaks. I hope you'll find the same on your journey. The best part is now that I'm retired, I'm back to programming in 8-bit assembler and 4K of RAM, except now I have a rather expansive collection with every model Atari, Commodore, Sinclair and Amstrad ever sold. :-)

Teaching myself assembler was harder because I had no docs. I was so clueless I actually looked up Motorola in the phone book and called to ask where I could get a manual. The sales rep who answered the phone must have taken pity on me because he mailed the CPU manual for free. I still have the purple reference card and remember the common opcodes by heart (https://colorcomputerarchive.com/repo/Documents/Datasheets/M...)

The first time I made money was about 18 months later when I wrote a pretty handy disk utility program. Stayed up all Friday night finishing it. Copied onto floppies I sold for $5 at the local user's group Saturday morning. Made $140 and got my first 28 customers. That was enough to get me hooked on making software and selling it. ~40 years, three startups, over a dozen products, hundreds of good people employed, a couple million customers and two PC Magazine Product of the Year awards later - I regret nothing. Started on a lark to learn and have fun. Much was learned, fun was had. Wonderful journey - highly recommend.

I agree he may have been thinking that, but it's another sign this guy was not the sharpest tool in the shed. I'm not an expert myself but was once stranded in an airport with an F500 CFO who definitely was. That's when I learned it's surprisingly difficult to make millions of dollars trading on insider information. While you can definitely make the trades, the hard parts are:

1. Getting away with it. Making a lot of money quickly requires being right when almost everyone else was wrong, which is easy for watchdog algorithms to spot. And if you work for a company involved in the transaction you're already on a watch list, so you have to enlist someone else you really trust as your accomplice to make the trades in their name - and it really helps if they're someone who has an existing history of making this size and kind of trade. Otherwise, your accomplice's name will automatically start at the top of the suspect list for investigation.

2. You need the right kind of actionable opportunity to come along. Transactions for which insider info could let you double or triple your money in one deal are pretty rare. More than that usually requires leveraging yourself up or having a lot of cash to put at risk. Unfortunately, the nature of insider info is it's about transactions which can end up being delayed or not happening at all at the very last minute for random reasons. So 'going big' with cash or leverage usually comes with a >~10% chance you could lose that cash or whatever assets you borrowed against.

Yes, it's potentially possible for an expert to find or engineer exactly the right kind of oppty, already have the resources required, and to have already prepared the mechanisms needed to get away with it undetected. But if you're that person, you already know all this. This CFO had been at the nexus of insider info for 14 years and could only think of one instance where he might have made big money trading on info he had, and that one blew up 6 hours before the announcement when the CEO of the acquisition target was randomly arrested in a prostitution sting (partying too early I suppose).

Conversely, when they award $2.00 of damages, to me they're telling Qualcomm "You're technically right but shouldn't have wasted everyone's time bringing this stupid lawsuit."

Press releases from large corporations announcing the FBI arrest of an employee under allegations of potential industrial espionage follow a standard format: dry, brief and to the point. They provide little context and volunteer few details for very good reasons. Prosecutors haven't filed official charges yet and it's likely going to criminal trial. In a couple days the accused will have a defense attorney who'll be examining everything that's been publicly said because it's now all discoverable and potentially prejudicial, opening up OAI executives to super-fun depositions. But... none of the competent PR professionals involved have a VP title and got overruled by an amateur who does (and that amateur apparently feels needy about highlighting in a brief, sober arrest report how the company's product was so, very useful in a national security investigation - because apparently neither OAI security nor the FBI could read the Seeed Studio logo and part number printed right on the device and, y'know, Google the product listing).

While the headlines and quotes seem designed to imply sophisticated industrial espionage, potentially by a state-level actor, what's public so far isn't adding up that way.

1. You're not going to get much actionable research material on proprietary technologies by audio bugging what's said in an executive conference room. That's what you'd do if you were after economic or deal negotiation info.

2. Hard to imagine a state-level actor buying a ~$15-$35 hobbyist ESP32 board from Seeed Studio (I've ordered a couple from there for home automation myself) and then asking a public chatbot for set up instructions.

3. They say he was an "engineer" who'd been there ~15 months. If you're an actual engineer at OAI and past the 1 year cliff, you're probably already sitting on RSUs likely to be worth at least $1M after OAI's IPO - and maybe much more. That's a lot of incentive to not get fired and probably makes you pretty hard to recruit as an internal mole for cash.

Yes, they did. The people who felt it was totally normal to add that would no doubt be surprised how desperate and cheap it makes them look.

Gleaning some semblance of nuanced, balanced reality has never been easy but in recent years it feels much harder. None of the handful political perspectives, media outlets or public policy institutions I felt I could usually trust are still reliably trustworthy. They've either been partially co-opted by special interests or have just grown muddled and conflicted on some issues. Even the ACLU, an org which for decades was so principled on free speech it marched in support of the KKK's right to speak, now feels some legal speech is too 'problematic' to be heard. Staying rigorously consistent now requires questioning everything, especially things specifically designed to appeal to my biases or emotional sensibilities.