The first stage of the revolution is to keep a cordial relationship with the Mensheviks. We're all on the same team. We're hear to overthrow that rotting edifice of the old order and create a stronger, better society, with a government truly representative of the people. We're all a big tent, and can accommodate differences of opinion.

The second stage of the revolution requires the sidelining of the Mensheviks. Yes, they have their objections, but those objections are mainly spurious, perhaps a little too influenced by moderate opinions. It's understandable, revolutions have casualties, and not everyone has the stomach for the hard fight. Objections will be duly noted and recorded.

The third stage of the revolution requires the expulsion of the Mensheviks. They've become too influenced by counterrevolutionary ideas. The middle ground they try to occupy is the path back to the old order. The revolution cannot afford these divisions, the people must see unity lest they question the revolution. Show the counterrevolutionaries the door, we no longer recognize their standing.

The fourth stage requires the destruction of the Mensheviks. It is not enough that they have been rendered impotent, they are traitors to the revolution, and like the moderates, in the hands of the old order. Some, maybe, can be rehabilitated, others must face more severe punishments. We owe to the people to destroy those who would undo our accomplishments.

The fifth stage has no memory of the Mensheviks at all.

I think you need to provide your operating definitions for "socialism" and "communism".

Rather ironic that the real Communists are the Republicans.

Well, some of that is for classes for people who can't see that default 3-pixel wide scrollbar on Windows 11 in high contrast dark mode. :-)

Yeah, you're blessed to have one of each. Until they start conspiring against you, which you KNOW is going to happen.

For all my pro-life ramblings, we were granted only one child. He's 22 now. I'm not sure if I'm jealous or worried about you. Keeping up with two toddlers after age 50 can't be easy.

In other news, by marrying close to my age, I'm spending 51 to 54 in caregiving activities to the point that I can't keep up with a 40 hour a week job, and the job market is such ain't nobody gonna hire me anyway, so I'm striking out on my own and trying to trick old people into paying me $300 to back up their Windows 10 boxes and switch on TPM in UEFI if it exists:

https://ancillary-proxy.atarimworker.io?url=https%3A%2F%2Finformationr.us%2F

I've been busily installing Debian on our Windows 10 machines. A few little hiccups here and there, but all in all, it's working like a dream.

The only Windows I use is the Server 2016 RDP managed service my company pays for, so updates are invisible to me. My two MacBooks and my Ubuntu laptop all have sane update policies which remind me of updates, without endlessly clogging up the works by downloading the updates. Every time I use an actual Windows machine I'm reminded of what an appallingly bothersome workflow-interrupting OS it has become.

It is literally a summarization model. Its context window is the text to summarize. It was literally trained on being given text as a context, and summaries as outputs.

Yup. I remember going to an IBM seminar around 1994 or 1995 where they demonstrated a new IDE environment that was going to end traditional programming. They gave a demo of writing some sort of simple application with input, with a library of GUI windows connected via some sort of flow chart. At the time I thought "Fuck me, I'm out of a job", but I never really saw the product again (for some reason I think it used Smalltalk, but it has been thirty years) and when I started using visual tools, it definitely wasn't the connect-a-dot that everyone claimed.

Worse, the stuff that was connect-a-dot, like all those horrible MS-Access applications written with Visual Basic, or the insane Excel sheets using lookups to make spreadsheets behave like RDBMSs, if RDBMSs had been written by victims of errant brain surgery, my career quickly morphed into a series of contracts in the vein of "Please fix the awful system we built in-house and we run all our Accounts Receivable through, but the guy who maintained it got hit by a bus."

(Which isn't actually much of an exaggeration, I had to take over a PHP project that had been half written by a guy who got some sort of serious illness, was taken over by some other guy who had no idea what he was doing, and the company had already sunk $40k into).

AI Overview literally tells you what sources it's drawing on. I don't see why people are capable of looking at what the sources are when they're Google search answers but not when they're AI Overview source links, but apparently, that is hard for some people.

This. Most people inevitably respond in these threads talking about "the model's training". AI Overview isn't like something like ChatGPT. It's a minuscule summarization model. It's not tasked to "know" anything - it's only tasked to sum up what the top search results say. In the case of the "glue on pizza" thing, one of the top search results was an old Reddit thread where a troll advised that. AI overview literally tells you what links it's drawing on.

Don't get me wrong, there's still many reasons why AI overview is a terrible idea.

1) It does nothing to assess for trolling. AI models absolutely can do that, they just have not.
2) It does nothing to assess for misinfo. AI models absolutely can do that, they just have not.
3) It does nothing to assess for scams. AI models absolutely can do that, they just have not.

And the reason the have not is that they need to run AI Overview hundreds of thousands of times per second, so they want the most absolutely barebones lightweight model imaginable. You could run their model on a cell phone it's so small.

Bad information on the internet is the main source of errors, like 95% of them. But there are two other types of mistakes as well:

4) The model isn't reading web pages in the same way that humans see them, and this can lead to misinterpreted information. For example, perhaps when rendered, there's a headline "Rape charges filed against local man", and below it a photo of a press conference with a caption "District Attorney John Smith", and then below that an article about the charges without mentioning the man's name. The model might get fed: "Rape charges filed against local man District Attorney John Smith", and report John Smith as a sex offender.

5) The model might well just screw up in its summarization. It is, after all, as miniscule as possible.

I personally find deploying a model with these weaknesses to be a fundamentally stupid idea. You *have* to assess sources, you *can't* have a nontrivial error rate in summarizations, etc. Otherwise you're just creating annoyance and net harm. But it's also important for people to understand what the errors actually are. None of these errors have anything to do with "what's in the model's training data". The model's training data is just random pieces of text followed by summaries of said text.

I remember when IBM, SGI, Infornix, Oracle, and HP first got involved in Linux. At the time, I included patches from some of them in the Functionally Overloaded Linux Kernel.

I proposed, back then, a simple league table for commercial support of Linux: Every new major feature or software product got so many points, and every bugfix release got a smaller number of points. Kernel features that made it into the mainstream kernel would qualify as goals for, kernel features and products discontinued were goals against. Closed-source contributions got half points, and were also considered goals against.

It would then be obvious which companies were serious and which were piggybacking, and it would also be clear who understood the philosophy, not just the opportunity.

Such a table would have ensured that nobody forgot the companies who contributed. Quite the opposite. There'd be an incentive to encourage the team you supported to improve position in the table.

Of course, no such league table ever happened. I could have maintained such a table without difficulty, but it would require the vendors to openly say what they'd contributed. I couldn't invent one out of thin air.

So I'd say Oracle has to look at themselves, not just the Linux community.

It's possible to conjecture - we know it collided with something massive, so if said body contained very limited radioactive materials, one might expect this to reduce the radioactivity per unit mass.

Is this the answer? Probably not, but it's good enough (I think) to argue that a simple answer is possible.