> Models are getting 99% more efficient every 3 years - to get the same amount of output, combined with hardware and (mostly) software upgrades - you can use 99% less power.
This is such a poor argument for a number of reasons.
1. Three years ago is basically when the "AI race" really kicked off amongst the frontier companies. You're effectively comparing a car from the 1920/30's to a modern car.
2. Past performance is not an indicator of future performance. You can't just say that LLM's will grow and improve at a fixed rate for all time, that isn't how they or anything else works in the real world.
3. Since it's an open secret that companies like Anthropic and OpenAI are running their models at a loss, a static 99% cheaper every three years arc still puts these companies at a net negative position unless compute, energy and water all somehow start getting 99% cheaper every three years.
I realize that the example is contrived, but what is the point of writing a test of a fibonacci function if your test harness is designed to just take whatever it tells you and updates the assert to verify that what it told you is indeed what it just told you.
This assumes the code you wrote is already correct and giving the correct answer, so why bother writing tests? If, however you accept that you may have got it wrong, figure out the expected outcome through some reliable means (in this case, dig out your old TI-89), get the result and write your test to assert against a known correct value.
I wouldn't trust any tests that are written this way.
Oftentimes, the main purpose of writing the tests is to prevent future regressions. This is pretty common for instance in ML (pytorch) code. You put in some tests with various random inputs and assert it against whatever your network says the output is. That way you don't accidentally change how the network works in future refactors.
These are great to test code that has no formal specification or oracle, so you're writing a reference implementation.
First, the test fails because there's no expected output, and you get to check the existing behaviour pretty-printed. Then, if it's correct, you approve it by promoting the diff into the source code, and it becomes a regression test.
Hmm, yes, I know that one and one of the reasons why I considered the naming weird. Original Tcl expect is rather automation than testing in the slot in my mind it occupies, even if it maybe could be (badly) used as a testing tool.
I believe it's called test-driven development but often I write tests hoping that what I tell myself the application code will do does what I want it to do. It's also self-describing of the changes made, and what people new to the codebase should reference if they actually want to learn what's going on.
And that's why the example is good. As you first get fib(3) then see the result, verify it and then freeze and then do for fib(20). That's how software development works that you can spot errors easier than a test could.
I'm willing to admit I may be the odd one out here, but I find this entire article to paint the author as self-entitled and manipulative.
> One might even call the art of accepting generosity a type of compassion.
People may feel good about themselves after performing an act of kindness, but this sentence makes it sound like the author is gifting them the opportunity to go out of their way to do things for him.
And all of these stories of wandering around Asia for eight years sound more like he deliberately put himself in positions to guilt people who culturally feel obligated towards generosity. These don't read like stories of kindness to me, but someone bragging about all the people he manipulated, and then recycled all these stories of manipulation into some ridiculous idea that he is the truly compassionate and kind person because he is accepting their kindness.
So not only are we measuring lines of code as a productivity metric as though that has any actual relation to productivity, but across the board they are boasting that lines of code is going up and PR density is getting bigger as well.
Those numbers should be seen as a giant red flag, not as any kind of positive.
I understand the intention of what the author is trying to achieve, but I think the problem they will run into is how do you define "evil" in a legal document or license? There is a subset of acts and beliefs that wider society has deemed "evil", but I doubt large corporations are actively supporting sexual assault, torture, murder etc. What the author is referring to is things they find morally reprehensible but do not reach the level of the aforementioned acts enough to be expressly illegal and evil (and whether they are or not, IANAL).
Several people either in this circuit or close by made submissions to this effect to ICANN recently.
It's very hard to get traction on this story because there is a lot of "don't prod the bear" regarding things ICANN can and should ask Department of State about, and things which really have moved into "self managed, independent international body" space. The reason there are two HSM east and west coast was because of this kind of national-strategic sensitivity. It would be a low bar (only money) decision to duplicate the investment in Singapore and Geneva, two locations which ICANN has existing investment in, with good secure facilities and accepted by the wider public as "neutral" points.
When the KSK ceremonies started up, several people also pointed out that this "diverse locations" thing was a bit hokey. The response above is my re-write of the kinds of things said to me, at the time. If somebody wants to deny State or any other US federal agency influenced the decision I have no formal proof.
I should add as a declaration of interest I was at Rob's goodbye KSK event, I am a TCR, and I made such a submission this year. I have not received any indication it was understood or read, despite asking for some acknowledgement, but the process wheels in an agency like ICANN run to their own time.
The risk is being told no, and inviting dissent into the independence of ICANN. Not asking, means no risk of being told "no, you do as you're told" which would endanger the whole 3 legged stool. the GAC would immediately question the assumption the US government had that level of signoff, the money flows and lawyers would fire up, it would be come a shitstorm in a teacup.
The least likely outcome of asking the department of state if ICANN is "permitted" to add an HSM outside the USA, is a positive answer.
The most likely path to doing it, is not to assume you have to ask.
It's my personal opinion from beer convos with people in the circuit. As I said I have no firm proofs and you should hedge belief in this by the lack of verifyable facts.
Yes, but we're a long way down "our hands are off it's ICANN now". The exception might be DNSSEC and the verisign contract continuance. I have no complaint against verisign, far from it: their staff are excellent and they are amazingly diligent and risk averse.
But at a contractual level you could ask is there another company which could tender to operate the root publication function, and meet all stakeholder requirements? And, could that company be legally constituted outside the USA?
Possibly. Ex CERN staff have indicated they were dismayed when the address management function went elsewhere in Europe. I know people both sides of this divide, it's ancient history in some ways.
Asking the US Dept. of State would almost certainly result in "huh?" from the folks there. The part of the USG that plays in the ICANN kiddie pool is US Dept. of Commerce (NTIA) and they no longer have a veto on what ICANN does.
One of the issues is section 4.2 of the IANA Naming Functions contract:
"[...] Contractor must be able to demonstrate that all primary operations and systems will remain within the United States (including the District of Columbia). [...]"
The Key Management Facilities are considered a part of the "primary operations and systems". IIRC, this clause was included in order to move the transition of the IANA functions forward in the face of some resistance within the US government.
Until that bit of legalese is revised, there will be no movement on creating a non-US key management facility. I believe changing the IANA Functions contract requires the Customer Standing Committee. As far as I am aware, no one within the CSC thought it worth the effort, i.e., "if it ain't broke, don't fix it".
Perhaps under the current US administration, that feeling as changed, but I haven't heard of any significant efforts in that regard.
The USA has shown, over the last 12 months, what a security-conscious country it is. The Defense Secretary's almost fanantical regard for messaging security should be held up as an object lesson for all future generations.
The problem with investors these days is that too many of them seem so focused on short-term gains and showing their numbers increase quarter-over-quarter that we seem to be incapable of looking at an investment over a longer period of time.
As someone who was the victim of a lot of bullying myself, this article is a very mixed bag for me.
For one, I like the idea of creating some degree of systems of support to try and prevent things like school shootings from happening by stopping them before they get too far.
On the other hand, unless there are more details missing from this article, it really seems like the only person who got any degree of punishment is the student who was being bullied.
You know what stops bullies that doesn't involve shooting them? Ruthless consequences for their actions. Schools love to talk about their 'zero tolerance' policy for bullying, but if there are no consequences outside of a teacher telling the bully to stop, then that is definitely less of a 'zero tolerance' policy and more of a 'mostly tolerated' policy. Zero tolerance means immediate suspensions, expulsions, supporting police reports for physically violent bullying etc.
Similar perspective from me, I really don’t understand why people in authority are allergic to making the correct decision and punishing bullies or even worse punishing the victim.
From my own personal experience being bullied. I went to teachers and the principal to speak up that I was being bullied, the teachers themselves witnessed it many times and acknowledged it was happening but the bully suffered no consequences other than being told to stop. My parents were awesome and got involved but even after that the school refused to do anything because the bully was “sorry”. Finally my parents told me they had my back and would support me if I wanted fight back but either way they were taking me out of that school at the end of the year. Punched the guy right in the face the next time after repeatedly telling him I would if he didn’t stop. I was immediately physically escorted to the principal’s office and my parents had to pick me up. The only reason I didn’t get expelled is because the bully didn’t want to admit I got the better of him so the school saw no fault. Never got bullied again by that kid. My story isn’t to say resorting to violence is the right thing to do but instead why did it even have to get to that point to begin with? So many members of authority could have issued consequences for behavior they witnessed but chose not to.
As someone who also was bullied heavily as a kid, my best explanation is that a lot of the adults who are in the posittions of authority were probably never the ones being bullied as kids (and some of them maybe were themselves the bullies). A large number of kids aren't directly involved in the bullying learn to keep their heads down and not get involved. The victims of the bullying will always remember it quite vividly, but those who just saw it happen without the same strong emotions attached to the memories won't recall the specifics of just how frequent or severe it was, but just have a vague recollection of bullying happening sometimes. When they end up seeing something similar happening in front of them again as adults, it wouldn't shock me if the same instincts around not getting involved or thinking about it too hard resurface and make it easier for them to rationalize not intervening.
For those of us who identify with the victims, this is almost unfathomable, but over the years I've been able to recognize that quite a lot of people don't actually identify in the victim in this situation. The idea that this might be the case didn't ever occur to me for years because of how much my insecurity and anxiety as an adult are related to my experiences of being bullied as a kid, so it made it hard to realize that this core emotional experience that's impossible for me to separate from my conception of what it's like to be a kid just doesn't exist for most people.
One of my boys was invited out by some classmates, then beaten up in a back alley. I called the police who visited their homes. They got the message that they'd be in court if they did anything like that again.
End of problem.
If your kid is bullied, call the police. Most school authorities are bully enablers.
Yes, the literal hours of video of footage of abuse that by the own claims of the authorities could have culminated in a violent response were "punished" with a single verbal warning. As the parent commenter says, they didn't get punished.
Ragging (esp. in STEM places; often violent; no slapping, et cetera, was not even considered violent) was a pathetic menace where I live. It is still not eradicated, but has shrunk to something so little that it is not a norm anymore or a right passage for seniors. How did it shrink? When seniors started getting expelled – no quarters given if ragging was proven.
Yup, cracking the toughest of entrance exams here after toiling for years in school (sometimes after) and going to those colleges and then getting kicked out (often with a piece of paper that ensured you didn't get admission elsewhere either) just because you couldn't resist harassing/abusing/attacking/hurting freshers who had just entered college did the trick. Before that? Threats, warnings, and policy-making just on paper did zilch. It was literally a national move sort of - coming right from the top, forcing states to act.. etc.
Is this sort of thing as common in US colleges as it is portrayed in movies?
My experience of university (not in the US) was that by then students had grown up, and there wasn’t any bullying going on that I saw. Students were treated as adults, violence was dealt with by the police.
Why would a bully care if they are suspended or expelled.
The most stubborn bullies in reality will often only reliably respond because they will starve to death, face violence, or have necessities taken away. I.e. a Nazi can't do nazi shit at work or they will get fired and starve.
A child cant get fired. Their parents must provide no matter what, and it is neglect if they don't and abuse if they use violence. End result is a bully knows the worst can happen is they lose luxuries and get a vacation from school, but always be taken care of. So really any punishment you can mete out is a nothingburger.
It's also very different to be warned that you "could get suspended or expelled" and actually have it happen. A warning isn't a punishment, but a communication that a punishment might occur, assuming that the one giving the warning actually follows through with observing whether behavior changes and is willing to actually carry out the punishment in the case that it doesn't. Kids are just as aware of this as adults and can make judgments about how likely this is when they receive warnings.
In fact, a child can get fired. They can get kicked off of sports teams, …. But yes, it is a difficult problem. I liked a lot of what I read in the article.
First, they identified the problem students.
Second, they tried something.
reply