...when officials filter out the AI by demanding handwritten submissions.

Few programs in the world are as exhaustively documented as Excel; there are so many doorstop books cataloguing the use and output of every object, method, and function.

So, that's perfect for reverse-engineering it, as long as you have a lot of programming time - or programming suddenly got almost-free. Just shove all those Excel manuals into your best coding LLM and tell it to clone excel with free code we can all use.

I will finally be impressed with AI-coding; and I will finally be able to leave Microsoft entirely, because the giant mountain of tools that is excel is my one reason I just can't leave; it's too useful.

Oh, clone ODBC while you're at it. Database hits are what Libre Calc particularly sucks at, and for me, database-front-ending and reporting was it's finest feature.

All over the continent, Emergency Rooms are backlogged and overworked. Doctors claim that paperwork steals a lot of their time, drags their efficiency.
And "medical paperwork" isn't about re-inventing a company, or building a new factory, or anything else that requires creativity and New Ideas: it's mostly the same 100-odd tasks over and over, the most, ah 'popular' diseases and injuries.

So doctors should be able to just do their medical treatments in an age of AI, narrate their day into a microphone as they go, and all the paperwork should just be done by itself, freeing up, what, 20-30% of all doctor time in the medical system? As if hundreds of thousands of doctors just popped into existence to clear away all the backlogs, give better treatment?

So: this is just about the easiest paperwork job to automate, and surely the most desperately-needed top priority. If AI doing white collar jobs isn't bullshit, then we should be seeing a major improvment in ER wait times, this year. Next? When? Please be specific.

In "Accidental Empires" pen-name Robert X. Cringely wrote a history of the first wave of microcomputer development - mostly Jobs and Gates and Aldus/Adobe, etc. In the first chapter, he points out that new information technologies

1) take 30 years to sink in: telephone invented 1870, changed lives after 1900; motion pictures invented 1890, big industry by 1920s; television invented 1920s, major industry by 1950s.

2) Rarely do what was envisioned at first. Bell thought phones would be used to broadcast music, and radio pioneers thought it would be used two way; both were opposite to each other. Early phonographs had record heads and they imagined spoken "letters" sent by mail...

Predicting WHAT the LLMs will actually do at this point is asking to be a comedy meme in a few years. People will just have to muck around with the tools for a while before they find out what they will really do for us.

Ironically, Cringely's point in that chapter is that the microcomputer revolution had to be invented by amateurs mucking around, because it takes too long for a new technology to settle in, to be worth the investment. He also notes that aviation was much-invented by the leftover planes and leftover pilots from WW1 just barnstorming and tinkering.

This time, the investors think they CAN stick around until payday comes...but if this one is also still 25 years away (I think we're at year 5 now?), then I don't see the money lasting that long.

Over-the-air HDTV is still putting out up to 18Mb/s per channel. It's some of the highest-quality streaming that there is, and it's free!

I've been recording it with entirely legal equipment for about five years now, use FFMPEG to crush the huge files down to H265 or H264 for action shows where movement shows some artifacts at H265. But mostly we just watch the shows within a few weeks, lots of room on the SSD of the ultrabook that controls the TV.

http://brander.ca/cordcutcuug

https://www.volts.wtf/p/whats-...

David Roberts of the Volts podcast asked full-time, full-career China expert Lauri Myllyvirta in April 2024, this exact question and got a clear answer: they're building them because they are forced. To get local permission for other projects, from the local regional boss (think "Duke") whom Xi needs to keep power. They can defy the national direction to some extent.

To build your solar/wind farm in China, you often have to build a coal plant, and buy coal, since the local Duke sells the stuff and hates the whole solar thing. So you get a lot of coal plants. What you don't get is more coal sales than they can get away from. The plants are often at very low capacity factor, sometimes under 20%.

Volts.wtf is strongly recommended for anybody wanting to keep up on the transition, the bad news as well as the good.

This might provide additional evidence for/against the thesis of the famous Rolling Stone article ("Criminal Element") by the late, great Kevin Drum, connecting crime to lead.

All that Drum had was the decline in crime after lead was removed; thing was, the lead levels and crime rates showed correlations down to state, county, and neighbourhood stats.

If we could also correlate with lead levels right in the residents, it would support the thesis, which still strikes many as too simplistic, crime being complicated, and lead levels, not.
https://lead.org.au/lanv13n2/l...

>Amazon doesn't pay bribes, ever.
Just $40M for the IP rights to "Melania"?

I was just adding up how far back GOP hostility goes. The Colbert quote is TWENTY YEARS OLD THIS YEAR.
It followed his previous sally "Truthiness", which was the word of the year for 2004.
Newt Gingrich killed off the Office of Technology Assessment back in 1995, it produced too much science for Congress.

Not to forget that in 1992, GHW Bush believed in Global Warming and wanted America off oil. I suspect the serious anti-science, anti-truth fight, against the Inconvenient Truth, dates to about 1993.

It's clear nobody clicked through to the MIT study. It actually is about running renewables 95% of the time, not 100%.
I should have not used the word "batteries", when I meant "storage".

Battery has always been a somewhat generic term for "storage", but is kind of triggering around slashdot. Still, it will be interesting to see what % of service can be provided in the global south is Form Energy's iron-air batteries work as advertised...with 120 hours of reaction time. That's as long as the longest dunkelflaute in Alberta so far. In southern Mexico, with solar on the west side and high Gulf wind on the east joined up, I'm not sure you would need more.

As Nancy Pelosi said, "we're capitalists"...we all still gravitate to the cheapest solution that meets standards, and were long-willing to put up with the health effects of coal, ignore the whole climate-thing, to get it cheaper.
David Roberts in Vox, in 2019, covered the MIT study that worked out just how very cheap batteries must get to make intermittent a baseload-replacement:
https://www.vox.com/energy-and... ...turned out to be $20/kWh for CAPEX, back in 2019. So that's $25/kWh in 2026.

Sodium-ion batteries are widely predicted to reach down to that price in a few years, less time than it takes to build a reactor. That's if Form Energy, now selling in test batches, doesn't get there first with their with "iron air" batteries.

If the batteries can be produced, and work in the field (they are so far, batteries are the hot new grid purchase for 2026), then both fossil and nuclear projects will fail to find investors. That's capitalism for you.

The only dodge around it is to appeal for government subsidy on some ideological grounds. That's what renewable champions had to do until their solution became cheaper; and it was a hard sell, let me tell you.

Bingo.
I asked a friend who started enthusiastically using AI for coding, used it happily for various business bits of writing, summaries, etc.

So I asked him if he had to give up one tool: "AI" (all of them) or "The spreadsheet", he thought for about 10 seconds and said, "AI" for sure: you're in and out of Excel all day long.

That was me, too. Excel was absolutely essential to my productivity as a data-slinger, managing real-word data into and back out of largish SQL databases. The ability to just refresh a pivot table from SQL was an automatic one-click updated report, with no code.

I could do a whole bunch of massaging of data from plain text files, notes, cut-and-paste from other applications - or I could do several Excel formulas and maybe a short macro, and process tens of thousands of records into the big database.

It was about far more than "modelling" it was a swiss army knife of data massaging, reformatting, and above all, data-cleaning.

And, yeah, I've tried to get the same work done in Libre Calc, and it's not even half-way there. It would be great if somebody could pour some real millions into Libre and take away Excel's lunch, but nobody is even talking about it.

$100/MWh isn't remotely competitive any more, mostly, but because of the "base-load" need, you might get that much. If it can't produce power cheaper than that, though, it won't fly.

Nuclear dreams are now in a race with batteries, basically - if batteries get down to $20/kWh as the CAPEX, keeping around enough batteries for a dunkelflaute every few years, starts to compete with $100/MWh of baseload.

And then there's geothermal, just a big question mark right now, but the chancers doing pilot plants are definitely aiming for less than $100/MWh - and for power that may not just be base-load, but have some storage capability so it can flex up more power at night, hold it back during the day. If they make that happen, nuclear has to beat them or die.

When I retired (10y) I was a whiz with Perl, had learned enough Python to know I could switch over easily, and was being told by Paul Graham that if people were too dumb to see that LISP was the ultimate language that had made his fortune, that Ruby had the same deep structure allowing the ultimate trick of self-modifying code and true compactness and elegance and all that stuff the Great Programming Languages all had to have for the most-elite work.

Of course, I didn't have to work any more, and I hate writing toy programs, and didn't have a problem that really required it, so at 10y, the O'Reilly Ruby book is dusty, and when I have something too hard for a bash script, it's still perl. Which still works.

But I was feeling guilty about it, and now I can put the Ruby book away with satisfaction that the moment passed. (Giving up on FORTH was the hard one; loved that language.)