NyxWulf

I think generally you can still rely on that aspect, compared to the alternative of UUIDv1-4. Even the worst cases he mentioned would cause minimal page splitting problems in a database compared to even the best cases with UUIDv1-4. The biggest take away I got from the article is that there is some wiggle room in how things can be implemented in the UUIDv7 spec, and it's important to be aware of the implementations you are using.

Many schemes do exactly this, if all you are doing is single table inserts that works pretty well.

If you do anything that creates related records via foreign keys you start to run into lock contention. For example:

If your insert requires a parent record with a single child record, you have to get the next sequential id which involves a non-preemptable single threaded lock, increment it, and send it back to the connection which requested it. In most databases this would mean writing the parent record and getting the sequential id of the parent record back so you can set the foreign key reference on the child record. Stored procedures are one way to mitigate the network transit time, but they don't solve the single threaded lock acquisition time.

The reason most distributed systems use UUIDs or another flavor of distributed id is so you can generate the id for the parent record on the client, set the foreign key there as well, then send both records in as simple inserts. It's not that there aren't ways to work around it or mitigate it. Using a coordination free id, is just much easier to work with and scale in practice.

In my testing the primary bottleneck isn't about writing to the same partition. Lock coordination to increment the serial key, then send it back across a connection was the real culprit. Virtually all databases make updates into RAM and to a write ahead log, which is then spooled to disk in the background. No matter which pages are written or updated, the changes all go sequentially in order into the WAL. For me, getting rid of Serial Keys was the big hot spot in processing. YMMV.

The biggest advantage I've seen for UUID come with distributed systems imo. The ability to create an id without lock coordination allows for much easier horizontal scaling. UUIDv1 avoids the collision issue by using a portion of the MAC in the id, but v1-4 all have sorting issues. UUIDv7 is a good choice if your use case warrants it. If you can get away with a single integer field like a database primary key and not get bottlenecked that's much easier. When that doesn't work anymore one of the distributed id schemes like UUID are the only real options available.

It's not just index pages. Generally databases write records into pages, which are then grouped into extents. If the primary key gives an ordering you can create new pages and start filling them up in an orderly fashion. If it's random, you are doing random inserts into pages everywhere all the time, then you start running out of room in pages and have to start splitting them. UUIDv7 solves the most painful parts of this problem in general and is a much better option for a Primary Key than UUIDv1-4.

He did say murder is well covered in criminal law for that, but things not leading to murder like these AI cases are not.

The writing sounds like Claude to me

Recursive Self Improvement

Which models, and at what effort level?

The author didn't argue that you should never use radians, or tau. Just that for some use cases there is an alternative that is simpler using turns. Your argument can be true and his proposal can also be true, it just depends what your use case looks like.

Here is an excerpt from the system prompt for UltraCode (Same for Fable,Opus,Sonnet):

"Ultracode. When a system-reminder confirms ultracode is on, that opt-in is standing: author and run a workflow for every substantive task by default. The goal is the most exhaustive, correct answer you can produce — token cost is not a constraint."

Maybe I'm not understanding this, but how is it that you can know enough about the data to process it without undermining the fundamental concept of encryption? Isn't encrypted data supposed to be just random noise without the key? The more you know about the underlying data the easier it gets to decrypt? Does this mean someone can just steal your encrypted data and use that to steal your identity without even needing to decrypt it anymore?

The arrest/conviction generally follow the pulled over stopped rates. It's pretty hard to arrest/convict people you aren't stopping. These types of biases and stereo types are often pervasive within certain groups of people, including law enforcement, prosecutors, judges, and often the subset of the local community in power. If you've ever been to court (even traffic court) you'll find the word of the officer is given tremendous weight, while the word of the defendant is almost totally marginalized (no matter what your race). Even though you are "presumed innocent", the defendant has a natural incentive to lie and the officer presumably doesn't.

To top that off there is almost no "suitable oversight" of any police departments in the US of any kind, let alone by the community.

There really are no simple solutions to this, even places not plagued by bigotry and racism have major structural power problems for the average citizen.

People believe a lot of things, especially when the other side has power. The point is that people complain about how the power is used by the other side. When power switches sides, they don't roll that power back. They pretty much always leave it in place.

Both can be true, because the experience depends on the skill of the user. The article the other day here on HN that LLMs reward skill is my exact experience. If you are are good at what you are trying to use it for they can be a skill amplifier, and they are definitely getting much better rapidly for the work I do with them. At the same time people are complaining that they are getting dumber. Saying that both can't be true ignores the skill requirement to use them and the completely different perspectives of people using them.

Even if both aren't true, your evidence was people saying two opposing things. The truth (if there is a single objective truth on a giving thing) has little bearing on whether or not different people agree on it.

It's an interesting point, but this is more like we are building a giant autonomous canon, that escaped the lab, the testing range, defeated state of the art and serious security protocols, and then blew a hole in the neighbors house.

Our legal and philosophical perspectives are deeply rooted in humans being the actors. Doing that in a residential home is unforgiveable. Doing it responsibly on a military range is expected. The autonomous agent escaping that containment then taking that danger somewhere unexpected and unprepared is something none of us or our legal systems are truly prepared to grapple with yet. Something which I think will require a reckoning sooner rather than later.

Ironically Hugging Face had to use a Chinese model to stop a Rogue US AI, since the Guard Rails prevented them from using Sol or Fable to remediate this attack. LOL

It's hard to overstate how important this point is. If China is the standard of openness, it's a pretty low standard. Your point I think adds to what the article is saying, from a different perspective, but arrives at a similar place. Our best selves in many ways are defined by openness, unflinching self reflection, and competition. We should remember that, and as the author of the article says, lean into it rather than letting fear mongering and histrionics protect these models from competition.

After having used LLMs for some time now, I don't agree with the concept they are just token generators, unless you think that's all humans are too. The way we test in most schools is just picking the right token. We also give them unique problems that they never saw in their training, which is the nature of programming. I realize they are probabilistic token generator models, but I find it harder and harder to accept that somehow there isn't something more going on. I'm not sure whether they are intelligent or not, but for the most part token generation is how you get degrees too. The thing is a parrot just says things it has already heard, it doesn't perform complex reasoning on novel situations and then explain it succinctly.

This is an interesting knot, the thing with knots though, you have to spend enough time with them to get familiar with tying them blindfolded from memory. My experience is most people don't care enough to do it.

So I have a simple alternative to tying my shoes that you can teach and learn easily. Knots are all about the number of turns or wraps, so when tying your shoes instead of crossing the laces over once, do it twice. When you wrap around the loop, do that twice too. You may have to try it to understand, but it is easy and readily understandable to anyone who can already tie their shoes. The best part is the way you tighten it down and untie are are exactly the same as you have always done. It almost never comes untied, but still releases easily.

When I was taking physics at the University 20 years ago, one of my professors said that the closest we have ever come to predicting an asteroid impact was "whoa that was close" as it zipped by.

Sounds like that's still state of the art.

My maternal grandfather changed his last name when he turned 18, from the German Uhren to Uhrey. If your last name bothers you, fix it.

This does cost money, but if you are serious about learning vim, this is an excellent game, and reasonably priced. One benefit I found is that the game creates a curriculum that exposes you to things you wouldn't otherwise try in your day to day, simply because you don't know them.

It helped improve my vim muscle memory substantially. Well done imo.

In the introduction to the study they include this statement, which I think agrees with your comment:

"This study systematically searched two databases and included six cohort studies included with limited types of digestive system cancers. Therefore, the evidence is not sufficiently strong to evaluate the relationship between digestive system cancers and plant-based diets. Comprehensive evaluations are scarce, especially for various digestive system cancers and multiple dietary patterns."

One of the features in go that makes it workable is you can assign it away via: _ = somevar

That eliminates the unused chain problem. Not sure if zig included that part of the feature as well.

Generally I agree though. The main problem with forcing unused variables, imo, is that it forces you to think a certain way. When I'm exploring the solution space on something, those constraints feel inhibiting.

A/B testing is a tool, and as Deming said, the aim defines the system. In your definition of Good, you are defining the aim.

I've been in a similar situation, where I created a relatively sophisticated A/* testing and control system. My idea of good use of the system ended up being very different from how the team employing the system thought about it.

I believe that is part of the point of the post, that unintended, and even unimagined side effects plague even the best of ideas.

I saw this when the Palisade Crest was the farthest point. Now there is a line almost straight down. It's been almost 15 minutes since an update, I hope that wasn't a tragic accident.

Can someone confirm if there is some type of speed descent down from the Palisade Crest?

Update: it rerendered, now it looks like a rendering artifact (I hope).

Did you read the article? The entire discussion was on the tension between immutable data structures and cognitive complexity on one side, and machine performance and mutability on the other side.

So the claim is...these highly trained pilots and the other sensor systems didn't realize that objects cause IR glare, but this guy did. Case closed.

So the anomaly that the carrier group tracked, then scrambled fighter jets for. The object that four different pilots saw visually, and was confirmed on multiple different sensors. That was camera glare?

The video is only one piece of the evidence. The pilots are trained, and other sensor systems confirmed what the camera was showing. This analysis is pretty flimsy.