First point: just because you can measure it, doesn't mean it is useful. The truly useful metrics typically require some significant analysis, which tends to make it much easier to "game" the numbers.

Case in point: unemployment rate. It's only people who unemployed and are actively looking for employment. If you've managed to get on disability, you are no longer looking, so you don't count against that metric. Many states were, at one point (are?), seeking to get their unemployed people on federally-paid disability so they wouldn't qualify as "unemployed" anymore, so the state wouldn't have to pay unemployment benefits. Yes, that's states getting people on the federal dole so they won't have to be on the state dole. And we wonder why Social Security is so badly under-funded. If you've gotten discouraged and you've given up looking, you don't count. There are a lot of single-income families where another of the household members would love to have an income but they haven't been able to find work, so they've settled into being a full-time housewife or Mr Mom. And there are plenty of people who genuinely need full-time employment, with benefits, but are only able to find part-time work (no benes). They may be "underemployed" but they aren't "unemployed." And there are no official "underemployment" metrics.

Second point: we are, too frequently, chasing the wrong metrics.

Case in point: GDP per capita. That doesn't really tell you anything, especially if the 1% and 0.1% are skewing the hell out of the averages. If the vast majority of people are earning jack, but a handful are really raking it in, it makes it look like everyone is doing ok.

A metric we genuinely need: ratio of median income to median living expense. Medians are less-likely to be skewed by outliers, so median not mean / average. How much do people earn, in the median? How much does it cost to live, in the median? If the first divided by the second is < 1.0, people genuinely cannot afford to live there. Compute that metric for the USA, each state, each major metro, etc. Publish the numbers. If this area has a ratio that's > 1.0 (people CAN actually afford to live there) but that area has a ratio < 1.0, people need to be moving.

In the last election, the Democrats kept talking about how the economy is doing well, based on averages (means), while Biden was in the White House. They lost because the majority of the USA is experiencing the medians, not the means, and the vast majority of the US population is seeing things getting worse, not holding steady or getting better. If we had more metrics like this, it would be plainly obvious when the economy is going the wrong direction.

I suspect that, for the vast majority of the USA, the ratio is < 1.0. And people are feeling it. Which is why, when Democrats talked about how well the economy was doing (the 1% and 0.1% are doing really well, skewing the averages, while everyone else is getting a very-raw deal), the vast majority looked at that and asked "what the hell world are you living in? it sure as hell isn't the world I'm in!" And voted for the opposite.

E-fuels will be so expensive that only supercar owners will be able to afford them. We're talking $20 per gallon or more. Nor do they solve the problem of local air pollution.

Somebody should ask the CEO of IBM how many he thinks they'll sell. I'm guessing maybe five...

I have an old copy of GReader running on my tablet. It pulls from Feedly. The vast majority of what it pulls get skimmed but all of my in-depth tech articles / news comes through there.

I've been wishing, for some time, for a new RSS tool. Something which can use regex to categorize stuff, or possibly using some kind of Bayesian filtering for same. Something which can group articles, based on said categorization; if I'm in the mood to read about DiMethyl Ether R&D, having a category for that and perusing multiple articles from same would be nice.

Something which can detect if these two articles, from different sources, are referencing the same source article and find some way of amalgamating them. Something which can detect when two sub-channels from a given source have cross-posted the same story and only show me one of them.

Something which can detect when this article is part 1 of a series, and this other one is part 2, etc. and categorize them together, such that it's easy to "binge-read" the entire series.

If someone knows of apps which do these things, please spill.

Another, related problem is that, even if you can read the software, can you find hardware which will run it?

There is an answer to that.

How long can you expect various types of media to last?

Here's a recent video about exactly that.

While he doesn't get too far into the weeds, how many of us have an old flash drive sitting around, or an SD card, which could have bit rot?

Assuming, of course, that you can make an axial motor which fits his idea, at the price he's quoting, there are still a couple problems with this.

Vehicles have sprung weight and unsprung weight. Sprung weight is weight which is insulated from the road by the springs and other suspension elements. Your engine and vehicle body are part of the sprung weight. Your wheels and brakes are part of the unsprung weight (no springs isolating them from the road).

For a smooth ride, you want your ratio of sprung : unsprung to be as high as possible. That way, the weight of the vehicle keeps the wheels stuck to the road and the wheels bouncing up and down on the road transmit less motion through the suspension to the body. Adding a motor to each hub is going to boost the unsprung weight, meaning you vehicle will have a rougher ride. That's problem #1.

You want as few moving parts, as possible, as part of the unsprung weight because they take a genuine beating driving over the road. Putting the motors in the unsprung weight means the motors will take a beating, which means they're going to have a shorter lifespan than you might hope. That's problem #2.

If you could remove a vehicle's differential and replace that with one containing one or more electric motors, such that the motors are part of the sprung weight, that would help to fix both issues. That, however, would add considerably to the weight and price tag, especially when you consider most front-wheel-drive vehicles already have a very crowded engine bay.

I admire his thought about retrofitting existing vehicles and turning them into Plug-In Hybrid Vehicles (PHEVs). I agree that, if 90+% of your daily driving could be done on electricity, you would seriously diminish the amount of fossil fuels needed. I agree that, if you are only adding 20 miles or so of range, you don't need such a large battery. I agree that retrofitting an existing vehicle into a PHEV, without spending 5 figures doing it, could go a LONG way toward diminishing fuel usage (mod your existing vehicle, rather than acquiring a completely different one). I agree with his motivations, and I've pondered what he suggests, some time ago. Unfortunately, I just don't see such an easy way to reach the goal.

I'd be happy with a plug-in hybrid but precious few of them are on the US market.

What I'd rather see is a trailer with a gasoline engine and a generator, such that I could use a lower-range EV around town but still hook up a trailer (rented, not owned) when I needed to do a road trip which was well beyond the range of my batteries.

And me without mod points today. Mod this up!

At a prior gig, I introduced the IT manager to a wiki. It was free software which ran on a Windows server (they were an MS shop). At first, he didn't see the value in it. I started putting stuff in there. He'd come ask me something. Have you checked the wiki? Uhh ... oh ... yeah ... there's that info. He quickly saw the value in it and started using it as well.

Fast forward a few months. We hired some more IT folks. We quickly discovered what was poorly- and well-organized. It became a rite of passage for new folks to hit the wiki to find stuff and, if something wasn't there, someone with more experience would help them write it up. New folks quickly came to see the value in it and contribute to it. Stuff got doc'ed, widely, and well-used.

Fast forward a few years. I'm at a new gig. I'm trying to learn my way around the infrastructure and the coding standards. Oh, go ask so-and-so about that. Are there no docs? Nah, just go ask that super-busy person who is juggling the jobs of three people.

We had a wiki (mediawiki, no less) but it wasn't getting used. I used it. I wrote stuff up. More new hires, interns, etc. They're being told "oh, check his wiki page; he's got a ton of stuff linked from there." They start doing the same. It's become a running joke. Can anyone tell me how to do such-and-such? Yeah, I have a page about that; try not to act surprised. After about a month, they're not surprised; they're laughing along with everyone else.

If only one person in the organization knows how to do that task, that task has Bus Factor = 1 (only one person needs to get hit by a bus, or otherwise rendered unavailable, before your organization suffers). A good wiki, where it's easy to write clear docs, and search them 'cuz plain text, can quickly take any task to Bus Factor = infinite. If you want your organization to thrive, you need the Bus Factor as high as possible on everything.

Anytime I have to wrestle with how to do something, I doc it on the wiki. If I need to do it again, next week, I might remember how to do it. After a couple months, it's good thing I doc'ed it. Eagleson's Law says that any code you wrote, but haven't touched for at least 6 months, might as well be written by someone else. It's been flushed from your short-term memory. If you have a mortgage and car payments, cut that number in half. If you have a significant other, cut it in half. For each kid, cut it in half. Pretty soon, you're lucky if you can keep stuff in short-term memory for a week at a time. Spending 5 minutes putting it on a wiki page, which you can find when you need it, will save you a lot of "wrestling."

Do any of their "prototypes" actually produce usable power yet? If the answer is "no", it's a science experiment, not a prototype.

a) the probability energy prices will go up in the next decades is pretty high

Even taking into account the need for additional transmission and storage, the numbers I've seen suggest that electricity prices in mostly-renewable grids will likely be similar to current levels, or come down from Russian-induced peaks in places dependent on imported natural gas. Energy costs will come down because expensive oil will be replaced by cheaper electricity, used more efficiently.

those plants will likely last more than the originally planned 60 years (based on what we expect from existing reactors in operation in the world)

The present value of power generated 60 years or more from now is close to zero applying any sensible discount rate.

Not a power grid engineer but my understanding is that as well as advanced inverters, you can use synchronous condensers to provide inertia in a grid with very high instantaneous renewables penetration. https://new.abb.com/motors-gen...

I recall reading the "Mother Earth News Handbook of Homemade Power" back in the 1970s. They had an extensive interview with an Indian guy named Ram Bux Singh, talking about "gobar gas," which was an Indian colloquial term for methane.

It also featured an in-depth article from an originally-British guy who ran a major pig farm in South Africa for 70 years, turning the manure into methane and running a modified diesel engine on it. You can read the interview here for more info.

Not only did his investment in the methane production equipment pay itself off (in the late 1950s) in the course of about six years, it also helped with another major problem found on pig farms: flies. Fly eggs in the manure went into the methane digesters and ... that was the end of them. The fly problem basically disappeared and the pigs were healthier as a result.

Let's assume for a minute that ChatGPT's known inability to stop itself BSing when it doesn't know the answer is effectively addressed and it becomes a truly useful answer tool. Now, how are you going to fund it?

My understanding is that each chatGPT answer requires a few cents worth of computation to produce. For somebody who needs multiple answers per day that's going to quickly add up. If there's one thing we've learned from the history of internet businesses, it's that barely usable and cheap almost always wins out over better and expensive.

Maybe there will be a subset of people who will be happy to pay (stock traders who want digestable answers RIGHT NOW come to mind), but it's not at all clear that the mass market would.

So just like real academics, then :)

More seriously, you're completely correct. ChatGPT gets things right a fair proportion of the time but if it doesn't know is often all too happy to generate plausible sounding BS, and there's no way to tell which is which.

Except that ChatGPT will all too often coherently and convincingly provide a wrong answer, and there's no way to tell because it doesn't cite its sources.

Yes, there will be plenty of wrong answers in Google's search results too, but at least you have a chance of judging the credibility of the answer before you use "rm -rf /" to free up disk space.