I have a son that's learning to read right now so I've got some first hand experience on this. This article is making a lot out of the contextual clues part of the method but consistently downplays or ignores that phonics is still part of what the kids are taught. It's a bit of a fallback, sure, but my son isn't being taught to skip words when he can't figure it out.

He's bringing home the kinds of books mentioned in the article. The sentence structure is pretty repetitive and when he comes across a word he doesn't know he tries to look at the picture to figure out what it is. Sometimes that works and he says the right word. Other times, like there's a picture of a bear and the word is "cub" (but I don't think my son knew what a bear cub was), he still falls back on "cuh uh buh" to figure it out.

So he still knows the relationship between letters and sounds. He just has some other tools in his belt as well. I can't say I find that especially concerning.

Eh, not really then. If you have some behavior in those 50 copy/pastes that needs to be deleted, you've got to delete it 50 times. That's not easier at all.

If every young person voted, the Republican party would collapse until it took a hard left turn. This is not a paradox.

I'm using the radar network for dispatch and priority for tie breaking/to make sure the resources are distributed evenly.

All my loading stations are simply called "Cargo Pickup" and all of my cargo trains go to any of them with an opening. Once there, the station reports on the red wire the ID of the train in the channel corresponding to the item being loaded (unless another train is already being reported by another station with the same items).

On the demand side, stations look for the ID on the item they need. They copy the ID into the green network on the channel corresponding to their station name. In the simple case, a station serving copper ore to copper smelters copies the train ID from copper on the red network to copper on the green network. But stations can also request multiple ingredients in which case they have some other symbol in their name besides copper ore. (Of course, here too the copying only happens if no other station is requesting a train on that same channel).

Back on the supply side, the station looks through all the IDs on the green network and sends the ones that match the waiting train to the train. The train uses the symbols to activate an interrupt to go to the corresponding station to deliver the goods.

I just set this up today. I haven't perfected it yet. One minor hiccup is handling the fact that you have no way to atomically access a channel. So two stations could request on the same channel at the same time, corrupting the ID. But that only happens if the stations are activated to make a demand on the exact same tick. It's not so much that it's a constant problem, it just bothers me that it could be.

You can also set each train stop's priority via circuit. I've been setting it based on how badly a train is needed at that station.

Because even if it winds up being a bad study, it still evokes a deeper, more important “truth.”

I'm being sarcastic but that's actually what's going on here.

I feel that they would at least change the framing instead of directly mirroring the OP. "Hating people for no reason but their race" is pretty clearly the definition of racism. Usually racists reframe their argument as actually being about criminality or at the very least some fear of cultural change.

I was seriously considering adding a /s but I was like nah, that's lame it should be obvious.

The real racists are the ones calling people racist just because they hate people for no reason but race.

This is not the same thing at all. Trump instituted a zero tolerance policy, separating any family caught crossing illegally with the stated intent to dissuade families from making the trip.

Normally (including under Biden) the government separates children from suspected human traffickers or members of gangs that engage in trafficking. This is not to deter families. It's to protect children - sending a child back to Mexico with a human trafficker is an abhorrent thing to do.

Stop carrying water for Trump.

Brother, if you can't even get a sizable chunk of people to join you now you sure as fuck aren't coming out of an armed revolution on top. There's no shortcut to going where you want to go. You gotta put in the work to convince people at the ground level.

It's like the dumbest version of the trolley problem where the tracks are reversed. You could do nothing and people will die. Or you could pull the lever (convince a bunch of people not to vote for Harris) and a lot more people will die but, hey, at least you can say you did something.

... any more so than society could – or should – force them to serve as a human tissue bank or to give up a kidney for the benefit of another.

This fact is why abortion restrictions are unethical period. In no other situation do we allow the government to force a person to give up parts of their body to keep someone else alive, even their own child. But most people aren't ready to hear that.

He doesn't need a plan. Half the voters don't care if he has a plan. Plans are for Democrats.

Imagine thinking that’s a great way to convince people you’re the right person for the job…

Worse, imagine how stupid you'd have to be to actually be convinced that he's the right person for the job. And then despair, because half the voters are that fucking stupid.

No mention of Gemini in their blog post on sge And their AI principles doc says

We acknowledge that large language models (LLMs) like those that power generative AI in Search have the potential to generate responses that seem to reflect opinions or emotions, since they have been trained on language that people use to reflect the human experience. We intentionally trained the models that power SGE to refrain from reflecting a persona. It is not designed to respond in the first person, for example, and we fine-tuned the model to provide objective, neutral responses that are corroborated with web results.

So a custom model.

When you use (read, view, listen to…) copyrighted material you’re subject to the licensing rules, no matter if it’s free (as in beer) or not.

You've got that backwards. Copyright protects the owner's right to distribution. Reading, viewing, listening to a work is never copyright infringement. Which is to say that making it publicly available is the owner exercising their rights.

This means that quoting more than what’s considered fair use is a violation of the license, for instance. In practice a human would not be able to quote exactly a 1000 words document just on the first read but “AI” can, thus infringing one of the licensing clauses.

Only on very specific circumstances, with some particular coaxing, can you get an AI to do this with certain works that are widely quoted throughout its training data. There may be some very small scale copyright violations that occur here but it's largely a technical hurdle that will be overcome before long (i.e. wholesale regurgitation isn't an actual goal of AI technology).

Some licensing on copyrighted material is also explicitly forbidding to use the full content by automated systems (once they were web crawlers for search engines)

Again, copyright doesn't govern how you're allowed to view a work. robots.txt is not a legally enforceable license. At best, the website owner may be able to restrict access via computer access abuse laws, but not copyright. And it would be completely irrelevant to the question of whether or not AI can train on non-internet data sets like books, movies, etc.

It wasn't Gemini, but the AI generated suggestions added to the top of Google search. But that AI was specifically trained to regurgitate and reference direct from websites, in an effort to minimize the amount of hallucinated answers.

Point is that accessing a website with an adblocker has never been considered a copyright violation.

a much stronger one would be to simply note all of the works with a Creative Commons “No Derivatives” license in the training data, since it is hard to argue that the model checkpoint isn’t derived from the training data.

Not really. First of all, creative commons strictly loosens the copyright restrictions on a work. The strongest license is actually no explicit license i.e. "All Rights Reserved." No derivatives is already included under full, default, copyright.

Second, derivative has a pretty strict legal definition. It's not enough to say that the derived work was created using a protected work, or even that the derived work couldn't exist without the protected work. Some examples: create a word cloud of your favorite book, analyze the tone of news article to help you trade stocks, produce an image containing the most prominent color in every frame of a movie, or create a search index of the words found on all websites on the internet. All of that is absolutely allowed under even the strictest of copyright protections.

Statistical analysis of copyrighted materials, as in training AI, easily clears that same bar.