Not learning from new input may be a feature. Back in 2016 Microsoft launched one that did, and after one day of talking on Twitter it sounded like 4chan.[1] If all input is believed equally, there's a problem.
Today's locked-down pre-trained models at least have some consistency.
Incredible to accomplish that in a day - it took the rest of the world another decade to make Twitter sound like 4chan, but thanks to Elon we got there in the end.
Twitter is like it always was. Unhinged leftists everywhere you look. Calls to erase Israel from people with Palestine and Trans flags in bio with these posts getting hundred thousand likes. The only difference now is that there are also unhinged right wingers (simply as a function of them not getting banned anymore).
I like it, it's entertaining. It reminds me of the old internet days. Wild west full of propaganda, but from all sides, not just the pre-approved western liberal one.
I don't want people like Tucker or Candace or Nick banned, I want to laugh at their nutty takes.
I want to laugh at "boomers" getting one-shotted by all the fake AI videos. I want to laugh at conspiracy theories about 6 fingers and coffee not being spilled.
I get the argument that weak minded parts of the population may take these things seriously, but the answer shouldn't be just "let's crack down and clean up everything unsightly", as that a) doesn't work in the long run, b) presents space for conformity based social contagions to run wild and c) goes against the concept of true democracy (which I like).
Mutation is a test for the test suite. The question is whether a change to the program is detected by the tests. If it's not, the test suite lacks coverage.
That's a high standard for test suites, and requires heavy testing of the obvious.
But if you actually can specify what the program is supposed to do, this can work. It's appropriate where the task is hard to do but easy to specify. A file system or a database can be specified in terms of large arrays. Most of the complexity of a file system is in performance and reliability. What it's supposed to do from the API perspective isn't that complicated. The same can be said for garbage collectors, databases, and other complex systems that do something that's conceptually simple but hard to do right.
Probably not going to help with a web page user interface. If you had a spec for what it was supposed to do, you'd have the design.
Same issue covered on HN a few weeks ago.[1] This one has more motor theory but less machine learning theory.
Too much gear reduction, and you can't back-drive or sense forces from the motor end. Too little gear reduction, and your motors are too bulky or too weak. Reflected inertia goes up as the square of the gear ratio, as the article points out, because the gear ratio gets you both coming and going. So high gear ratios really hurt.
Robots, like drones, need custom motors sized for the specific requirements of the joint. For a long time, the robotics industry was too tiny to get such custom motors engineered, and had to use motors designed for other purposes. This will become a non-problem as volume increases. Especially since 3-phase servomotor controllers, which drones need, are now small and cheap. They used to be the size of a paperback book or larger.
(I've been out of this for years. I've used hydraulic robots and R/C servo powered robots.
The newer machinery sucks a lot less.)
Reflected inertia does scale as the square of the gear ratio but it's a bit misleading unless you also consider the change in rotor inertia, which scales as a cube of the rotor radius (as the article points out).
The other side of the scaling laws say that motor torque scales as a square of air gap radius (roughly rotor radius), and output torque scales as linearly with gearing ratio.
When you balance these out, the reflected inertia depends on the inverse of power dissipated for a fixed output torque.
In an ideal world, your total reflected inertia is independent of the gearbox and largely depends on the motor fill factor and how hot you can run it.
There are more conspiracies. Here are some well-verified ones:
- Epstein and way too many important people.
- The big one from the 1970s onward to increase the return on capital by lowering living standards, the "Powell memorandum".[1] That's the founding document of the modern conservative movement.
- Facebook/Meta being behind schemes for age verification.[2]
I wouldn't say that Epstein is a vindication of conspiracy theories, at least not the "Bigfoot" type. Epstein was already in trouble with the law for trafficking over 20 years ago. The pedophilia in the Catholic church was known decades before that. It's shameful that these stories didn't get more attention sooner, but the general veracity of them wasn't in question.
The prototypical pedophilia conspiracy theory we didn't believe at all is the Comet Ping Pong one, which was appropriate.
> The pedophilia in the Catholic church was known decades before that.
Except the proportion of paedophile priests is about the same as the proportion of paedophiles in the general population. There are more paedophiles in schools and social services than in religious organisations - and there have even been more convictions of teachers and social workers, at least tin the UK. The reason you think of the Catholic Church this way is BECAUSE it got more media attention earlier than elsewhere. A surprising number of people the UK do not know about the biggest big paedophile scandal in the country, the Islington one, that was huge, and at least one politician who was responsible for the failure to investigate went on to have a successful career in politics (the only time it set back her career at all was when Blair wanted to make her minister for children there was a backlash)
> The reason you think of the Catholic Church this way is BECAUSE it got more media attention earlier than elsewhere.
I think what earned the Catholic Church their reputation was that the church was actively involved in protecting known pedophiles and repeatedly shuffling them around effectively giving them an endless supply of fresh victims while suppressing the voices of many of the children brave enough to come forward. The problem with the church isn't that pedos exist there, but that they've been very often supported and defended and their actions covered up in ways that don't (and often couldn't) happen in schools.
I believe you but is there a source that you can refer me to? I guess I just believed abuse was more common in the Catholic church but your post made me realize my impression had no facts behind it.
> Except the proportion of paedophile priests is about the same as the proportion of paedophiles in the general population.
I doubt you have any reliable statistics about this, given how many victims keep silent out of fear.
But in any case, the moral failure of the church was not the existence of individual abusers (which indeed can exist anywhere in society), but how on an institutional level known abusers were protected by the curch. Everyone who was part of the cover-up (which went all the way to the top) is complicit.
I think if 20 years ago you claimed that there was a global sex trafficking ring that procured young girls for elites, politicians, celebrities, and royalty, you'd be laughed off as a David Icke level conspiracist. These days it just seems obvious that that was going on.
Its not just a sex trafficking ring, its a corruption ring, and the corruption part of it is much bigger. It is what the arrests in the UK have been for. Given how senior some of the people in the UK are (Mandelson is a former cabinet minister, and a former European Commissioner, and was very influential even before he held those posts).
If they had not trafficked minors as well I wonder whether it would ever have been exposed. It makes me wonder what else is going on.
There are big differences between the historical notion of a courtesan and the underage women trafficked by Epstein. The main one being that courtesans played an established and visible (if unwritten) role in society.
"Bigfoot" isn't inherently a conspiracy theory. If you say that bigfoot exists, you're wrong, but not necessarily a conspiracy theorist. To be a conspiracy theorist, you also have to posit a grand conspiracy to conceal the existence of bigfoot.
If you posit a conspiracy that only involves a few people who could plausibly coordinate to conceal the truth, that's also not a grand conspiracy, and we don't call people conspiracy theorist for believing in regular, everyday criminal conspiracies.
It wasn't meant to be philosophical, it was meant to be practical. As a practical matter, you're wrong if you say that Bigfoot exists, or that the sun won't rise tomorrow.
> If you posit a conspiracy that only involves a few people who could plausibly coordinate to conceal the truth, that's also not a grand conspiracy, and we don't call people conspiracy theorist for believing in regular, everyday criminal conspiracies.
No, but we did call people conspiracy theorists for believing the thing Snowden subsequently showed to be real.
Not me, I didn't. That conspiracy was certainly pretty big, but there was also a ton of smaller leaks as you'd expect on a real conspiracy of that size, so you certainly wouldn't be called nuts for assuming NSA were spying on a lot they weren't supposed to.
Security state loyalists were not nearly as influential in online discourse back then, as they are now. Probably astroturfing, AI and algorithmic amplification plays a part in that.
Shelf life doesn't matter if you are firing them as quickly as you can make them, especially if you actually can make them as quickly as you need them because they're so simple.
Right. High-volume users can skip the thermal batteries with decades-long shelf life, and just spot-weld a few AAA batteries inside the weapon. Just stencil the thing "Best if used by DATE". Good for a year or two at least. Skip the anti-corrosion stuff and ship it in consumer-grade shrink wrap. Ukraine ships drones to the field in lightweight cardboard boxes, not rugged weapons containers.
Many US weapons are really old designs. The Patriot went into production in 1980. The Stinger went into service in 1981. There's been progress since then. Consumer-grade parts can do most of what's needed.
AAA batteries don't have the current. Li-Ion is too fussy and has a pretty high self-discharge.
Ukraine can afford the cardboard boxes because they are fighting in their own country. The US has an ocean to cross.
Ukraine can get away with short shelf life because they are at war right now. The US has to stockpile because the supply chain has to run at some capacity in peace time to be able to ramp up quickly when needed, and discarding the produced ammunition after a year would be incredibly wasteful.
Neither Ukraine nor Russia can defeat each others' air defence networks. The US has a lot of experience doing just that, while successfully defending against ballistic missiles. High tier capabilities matter.
The Patriot in 1980 is a very different system from the Patriot that is fielded today. Between PAC-2 and PAC-3, AN/MPQ-65A and LTAMDS it's a cutting edge air defence system. The progress is constantly incorporated.
The Stinger is a bit old, but mostly because the US doctrine has few uses for it. Regardless, NGSRI is coming.
But this is exactly the point: This approach allows for insurgents or parties subject to overwhelming but expensive force to strain the logistics and budgets of their opponent. This is something that would be far more costly for the US to counter.
That's roughly how the original Sidewinder worked.
The original concept was to reduce near-misses. If the pilot could get on the target's tail and aim at the engines, it usually got a hit. That was the same task as getting into firing position for guns. Hit rate about 8% in combat.
Later versions allowed launches from longer ranges and from off-angles.
reply