I was working on a twenty-year-old Java project which had very few unit tests last year. In a couple of days with Copilot I added tests for a couple of hundred classes; for the simple ones it generated the entire test and I just needed to run it, for more complex classes it needed more help since it didn't understand things like not trying to mock final classes. The tests also found a handful of bugs that had been in the code for years so I fixed them.

Last week I had to write a test for old C++ code which maps between internal and external error codes because the test didn't exist and we were changing the mappings so I didn't want to break any mappings which weren't supposed to change. So I was able to cut the switch statement from the mapping code and past into the test then have Copilot convert the switch into a map of inputs to outputs so I could test every possible mapping by looking up the result in the map or checking for the default code otherwise (it also wrote most of that part but didn't get it all correct).

There's a modest amount of Copilot code in the actual released software but Copilot does tend to screw up by calling methods which don't exist and things like that which have to be fixed. So it rarely produces a perfect result. I suspect it would work better if it was a project which had been developed using Copilot as maybe then it would have created those methods that it later expects to call.

Another thing it's good at is taking Java code and converting it from looping through a collection into using a stream instead. I swap languages all the time and can never remember all the different stream methods in Java so I just write it as a loop and then tell Copilot to turn it into a stream.

Mostly it's taken out a lot of the work that junior developers might have done in the past. So we're allowing senior devs to do more work in the same time but eliminating the jobs for people who would progress to being senior devs in the future.

Is that you, Al Gore? How're ya doin'?

Still pushing that Information Superhighway Driving License to keep drunks off the Information Superhighway?

The whole point of this system is to fail so the Normies will demand a government-issued Digital ID to prove they're allowed to watch Youtube videos.

The "Vaccine Passport" failed to create a persistent Digital ID, so now they've switched to "Save The Children From Pron".

And it seems crazy to ship CO2 to Norway when they could just pay Norway to suck CO2 out of the atmosphere instead. It's not like there's any difference at the end of the day, unless you happen to own a shipping company that can make $$$$ from transporting CO2 for no good reason.

Exactly. In the 90s we still used to try to optimize C code by using register variables and complex function structure that happened to suit the way the processor worked.

Then we stopped doing that because we realized the new compilers could optimize it a heck of a lot better.

Now we typically don't even write programs that generate machine code any more but feed everything into a VM that generates code on the fly.

I don't remember having to do any serious optimization for years, and it was mostly stuff like batching up messages so we weren't trying to process them one at a time with all the overhead of starting and stopping the processing operation each time.

The last I saw, 50% of peer-reviewed papers can't be reproduced. So a 40-60% funding cut is probably about right.

Interestingly I was talking at the weekend to a guy who used to work for a soda company here. He said the glass bottles used to be made and filled locally where he worked because they were heavy and difficult to transport, but switching to plastic allowed the company to centralize production in a city hundreds of miles away and ship them here instead. And, obviously, sack him and most of the other local workers.

But the MBAs got a boost to their stock options.

We're no longer allowed to bury plastic because it requires digging holes and digging holes causes Climate Change or something.

This is why we now ship the plastic to the third world so they can throw it into the ocean instead.

I think it will be difficult for the music industry to push 'this song was created by an AI in which 0.1% of the training data was a song we own' because that would pretty much destroy the AI industry which is based on being able to use other people's content without paying for it and the Tech billionaires are richer than the music billionaires. You're right though that the tech companies like Youtube might agree to block AI music 'because of unresolved legal questions' or some such nonsense.

Imagine the economic chaos if I could just tell my computer what kind of music I want to listen to and it could generate that music for me.

Think of the starving Music Executives!

Because The Mushc Industry must be able to continue rent-seeking and collecting most of the revenue that actual musicians create.

I know a guy who was a moderately successful musician in the 90s with a few songs near the top of the charts back then. He loves AI because he no longer needs a band and can create entire songs by himself.

That terrifies The Music Industry. Because its income is based on gatekeeping and rent-seeking.

And Apollo 13 cost NASA billions of dollars and came close to killing the crew because they screwed up the oxygen tank tests.

If SpaceX didn't test the tanks before filling them that's probably retarded. But the Apollo 13 oxygen tanks were fine until someone damaged them with improper testing. It's not inconceivable that someone damaged these tanks after testing and either didn't realize or decided the damage wasn't serious enough to matter.

17 Apollo missions but only 13 Saturn Vs flew (including Skylab).

One of the earliest Saturn V launches (maybe it was the first?) lost several engines but was considered a success because it still made it to some kind of usable orbit. Apollo 13 came within about second of disintegrating during launch, but fortunately an engine shut down and reduced the hazardous vibrations that were going to tear it apart. Skylab's engines overheated because the interstage didn't separate, but that was due to a piece of Skylab falling off and breaking stuff on the way down. I don't remember whether any of the other POGO problems before Apollo 13 came close to destroying the rocket, but they'd almost reached the end of the Apollo program before they fixed those problems.

NASA also blew up a lot of Saturn V engines in testing, to the point where some people didn't think they'd ever make it work.

So, yeah, NASA blew a lot of stuff up to get the Saturn V to work and it only claims a 100% success rate because of luck and ignoring things that didn't really work right.

Running some calculations at the PVWatts site, Colorado looks like a pretty good place for solar power. An 8kW system is allegedly only producing about a third less in winter than summer.

We're quite a bit further north and get a quarter to a third as much power in winter as summer. Assuming the panels aren't covered in snow.

You do realize that 12.5kWh of batteries will only power our moderately-sized house for 8-10 hours?

I'd like to run my next house off solar but based on my experiments over the last couple of years we'd need at least 60kWh of batteries to cover the cloudy days in winter and even then we'd need to limit non-essential power usage until the sun came back because some days we get negligible power from our existing panels for more than two days at a time.

That would only be about $20k worth of batteries these days if I bought regular server-rack packs but then I'd need to convince our insurers that they're not going to catch on fire and burn the house down. It's probably twice that much to buy fully-approved batteries here.

So $120k is probably about right for a system which can power the house and keep that power on in the depths of winter. And keep our insurers happy.