jaybeavers

You’re not wrong. Me too. I spend most of my waking hours using AI for new product development.

But can you bring it to market and turn it into a reliable revenue stream? And will this still be true in a few years when the duo or trio -opolies corner the AI market enough that the $200 subscriptions go away to be replaced by API pricing only?

I’m already seeing my subscription use slow to a crawl during peak hours so I now code before 8 am or after 6 pm. And turning on ‘Opus Fast’ with API pricing is already financially impossible for me as a small business.

Ironically, China may be the savior here, with open source models that may be good enough to ‘put the means of production in the hands of the working class’.

I mean, is QWEN a 4d chess game to replace capitalism with communism? Who knew (outside of the CCP central committee)?

I spent a few years studying the ‘startup economy’ after a 20 year career at MSFT that left me with some meager capital, including a half dozen Angel investments and running an Angel investment fund LLC (SPAC as they would call it today).

Never saw a startup succeed that didn’t come with substantial founder capital, across a couple hundred evaluations. Does it happen? Sure. Is it the model for new product generation that people with <$250k of personal liquid assets can bring a new product to market? Not really.

‘You got to have money to make money’ is a real thing.

Yeah, I know I’m posting on YComb’s forums, but if you think the capital YComb itself gives to ‘good ideas’ is enough to succeed, that’s silly. It’s the follow on investment capital that brings a new concept to market, and that comes with odious strings for people underendowed with their own capital.

E.g. startups are still a rich man’s game.

“I find hope in your acknowledgment of your own imminent devaluation.”

Gave me a chuckle. I think this is my personal ‘best sentence of the year’.

Great way to summarize how we will likely look back on 2026.

We’re likely looking at a new world where physical and mental work from humans holds much less value.

So if the world still generates value ( think AI inventing new drugs, robots planting and harvesting fields of corn ), who is able to make income? The people who have the capital to purchase and run the systems.

So our future world probably looks like a system where effort/knowledge/skill has little to no reward and capital (e.g. inheritance, passive income) has all the reward.

So, we can have a world where you are born rich or you live in abject poverty, or we can figure out something else.

You have to admit there is now some lovely schadenfreude to be had from the whole ‘Chinese free LLM companies be stealing our theft! Stop them!’ whining.

Or, perhaps, the companies forming the for profit LLMs could be made to pay a license fee for the copyrighted data they lifted from behind the paywall and then used to form a for profit entity with.

Nothing about banning technology. How about enforcing DMCA and then applying a penalty for the knowing theft rather than negotiating a license?

Flourishing life? Where is your data leading to the conclusion that AI models are leading to the world populace leading a ‘flourishing life’? Are taxes on revenue of companies like OpenAI somehow being collected and turned into a UBI and I just didn’t hear about it?

You are correct, the driving controls in AVP don’t use dwell, that is the wrong (and dangerous) approach. They use something more akin to hover activation.

It’s the hardware I designed coupling the power wheelchair to the AVP, and I’ve driven it myself.

Hi Will, thanks for the representation. And an actual link!

Yes, daily, as a virtual screen for my Macbook using a BLE mouse and keyboard paired to the Macbook not the AVP. Then I supplement the Mac with the occasional AVP side app using fingers and eye gaze control.

If only the AVP ran MacOS natively on the M2. Somewhat ironic that I have to park an M1 Macbook on a table next to my AVP, screen open so I can unlock it (requirement for the screen pairing).

I’m considering putting my Mac Studio into ‘never screen lock’ and then using a long range Logitech dongle for the wireless keyboard and mouse so I can use my AVP without an opened Macbook by my side all the time.

A big bonus for me was VisionOS 2 added lower latency screen mirroring (and I think foviated rendering of the mirrored screen). Prior to that I had to use a developer strap with a second usb cable to the Macbook to get no noticeable latency on the screen and cursor.

Strongly disagree. As a dev who’s been in the business since the days of the first gui apps and OSes, UI can be very straightforward, but we have designed modern systems such that they are immensely complex.

Styles and inheritance and asynchronicity have turned what could be a simple approach into an inscrutable problem.

I took. my wife’s Model X out for a moderately complicated drive last night and was very pleasantly surprised by the new FSD behavior.

Very significantly improved over the last few updates.

Having submitted a number of “product breaking bug reports” through official channels while using a paid support incident, can confirm Kafkaesque and a multi year support process.

Oh, I can get a person to “respond” to a problem with a paid support incident. But the person responding is clearly hired only based on their ability to respond to an English email, once a week, with either, “thank you, I will forward that to our engineering team” or “thank you for your patience, our engineers are still looking into that issue.”

Severity, impact, paid support, none of that matters. Heck, it’s impossible to tell if it’s an actual person or just a sophisticated shell script.

Spoken like someone who hasn’t put a match to gasoline.

Having recently used gas to help start a wood slashings pile on fire, gasoline is very explosive and dangerous.

Diesel will burn in place when lit. Gasoline vaporizes when let out if its container in ‘room temperature’ conditions and creates a vapor fireball when it finds a nearby ignition source.

Never use gasoline as an ‘accelerant’ to start something else on fire —- it’s way too dangerous.

So Avalonia is built on top of Skia which is Google’s portable 2D rendering engine.

So in theory it is as cross platform capable as Chrome/Electron without all the overhead that html brings. Brilliant play, imho.

Now, as long as Apple doesn’t manufacture a reason to ban apps based upon it…

Probably a much simpler answer. The programmers who work on a washing machine are probably inexperienced low wage disempowered help. And there’s an innocent rx/tx loop for a home grown notification system and noone on the team knows better or knows how to use a packet sniffer or perf analyzer.

Never attribute to malfeasance what incompetence can explain.

Put a packet sniffer on it. Bet it doesn’t use https, that costs extra for microcontrollers. Dollars to doughnuts it’s a teenager on a road trip.

Are we there yet? Are we there yet? Are we there yet?

This was my experience too. While it takes a good long while to get resolved, because my mistakes were accidental and I talked it through and followed the process, I was dealt with politely and given grace.

All in all a scary but good experience.

Thank you from someone begrudgingly paying thousands of Euros for IEC medical device standards pdfs!!!!

Windows NT was rock solid until Windows XP was released, but GPU performance was meh and Windows games generally didn't work well unless they were written to target NT. As I recall, there was one commercial graphical strategy game for NT at the time.

XP broke the subsystem isolation of NT in favor of enabling Windows non-NT types of drivers with higher levels of kernel/GPU efficiency due to less isolation. This also brought in the ability of the drivers to crash the kernel.

And that, my friends, was the end of rock solid Windows NT stability.

I was on the Windows team during the XP SP2 years. Stability was somewhat reachieved by an intense focus on defensive coding to detect rogue driver and the establishment of the WHQL, or Windows Hardware Quality Lab. WHQL was basically an investment in driver analysis tools and a moderately sized team inside MSFT who's sole job was to debug and fix other people's drivers.

It's not a good replacement for isolation though, and it requires sustained continuous effort by both MSFT and the windows hardware partners, which imho hasn't continued.

Is there a Windows client front end ide rather than a browser?

It's much easier to enable alternative input from say a .Net Framework WPF app...

Nice to hear from you, great work! As a lifelong dev who started by learning BASICA and who's first pro job was in Visual Basic, I love your thesis.

Nowadays I work with people living with disabilities and I come across kids who want to program but who cannot use a traditional keyboard and mouse. I was just discussing the need for a simplified ide and language syntax with full programming power and Small Basic came to mind.

Any chance this work will see new life outside of its 'side project' status?

Both of those are reasonably met on the Steam Deck today which is running Linux…

Now the Steam Deck has an x64 CPU, but I’m running my heavyweight production Windows apps today on a Macbook Pro M1 running Windows 11 Arm in a VM which in turn is running Windows x64 apps in Windows on Arm’s cpu emulator and its a decent experience.

Really heavy apps like Visual Studio (not Code) and Altium (ee cad) are noticably slower, but run juat fine if I tune the VM for ‘gaming/CAD’ setings which basically allocates 12 GB RAM and 4 arm cores to the Windows VM.

Steam Deck does things differently — they have a sufficiently large Win32 compat layer that they aren’t running Windows in a VM at all. Now They don’t run this api layer on top of a x64 over arm layer, but I dont see why that won’t happen somewhere sometime soon…

Valve/Steam has the ‘runs existing Win32 apps without Windows working really well in production and they have a financial engine (games storefront) which is funding the support and improvement of this layer.

Both Apple and Microsoft (and others) have ‘run x86 apps on arm’ running really well in production today.

If Apple really wanted to run Win32 natively, they could do it with much new engineering investment. If Valve decides that an Arm chip makes more sense than their AMD chip in the Steam Deck, they could do it too.

Maybe some third party disruptor leverages Valve’s work and uses one of the (much better than Win11) Linux shells and creates some hardware like the Mac Mini M2 or the MSFT Arm dev kit, we’d see a real contender for Desktop Linux with full Win32 support on Arm.

The real question to me is does anyone outside of Apple have a current gen Arm CPU/GPU that has the performance? My Surface Pro X with Qualcomm Snapdragon is considerably slower than my Macbook Pro M1 running Windows apps in Coherence mode on Parallels Desktop inside a Windows VM. Parallels works, but I end up burning 6-8gb of ram to windows vm overhead and have the overhead of managing dual OSes to make this frankenstack work…

Pardon, what part of ‘a home with electrical service works just fine’ fails to fit your definition of ‘a large scale distribution of energy’.

We have an electrical power grid, dude. And it works just fine for home BEV charging.

You think his writing is a good time, you should have seen him when a VP would say, “Are there any questions from the audience?”

He’d saunter up to the microphone, and you could see the VP’s face droop…

Good times, my man. Good times.

You may be surprised to learn that James’ brilliant humor and willingness to speak truth to power didn’t work out so well for him at Microsoft.

He’s at Harvard now, which I think is a step up from Hollywood.

I have three BEV chargers at my home now. i didnt need to change a thing to my “built in 2000 to code” home to do this, as it “came out of the box” w/ 200 amp service.

My chevy bolt will charge at 8a (conservative) or 12a (“fast”) at my preference. It takes maybe 3-4 hours over night to charge from my average driving day with a decent commute.

My car consumes, while charging for a few hours overnight, about the same amps as a good clothing iron or a beefy vacuum. If i wanted, i could delay charge so three EV charged overnight, never draeing power at the same time.

My next car will buffer power so solar cells can overcharge and supply back to the house. And it will be a backup power supply for storms and grid outages so I no longer need a standby generator.

Tell me again about how the grid isnt ready for my BEVs and why we should be pumping some liquid or gas into tanks?

While this is a common refrain, as driver on my 3rd EV “charging network needed” a bad, bad misunderstanding propagated by people who dont drive EVs.

I charge my EV at home. I’ve charged my EV on someone elses charge (because i needed it, not because I got a better parking space) maybe twice since i moved to an EV with a 180 mile range. Heck, even with my “first gen” EV with a 63 mile range, I charged at home 95ish percent of the time, and those days are behind us now.

$2k to run a 240v external outlet, $500 for a charge station, and you’re set. Apartment dwellers still have issues, but thats simply because govmt/demand hasnt insisted yet.

At this point I strongly distrust any 'breakthrough' article about research at MIT. After hundreds of these, I'm fairly convinced that any time a grad student pours liquid into a beaker, MIT's marketing department is out publishing the fact that Flubber has just been invented and we're all going to be saved by bouncy flying automobiles.

I mean, congrats on the great marketing department. But it's tiring to be disappointed over and over by the hype.

And how do you clean these optics?

Do they still provide 110% optical efficiency while dirty? Or damaged/scratched from frequent removal/reinstall for cleaning?

+1 We need the Singularity to make sense of the mess known as PDF.

But realistically, might this bring on the SkyNet attack if we abuse AI so?