HNHacker News
TopNewBestAskShowJobs

candiodari

4,236 karma · joined September 25, 2016

submissionscomments
candiodari··on Why the world cannot afford the rich
These are the authors of this paper:

https://wilkinsonpickett.com/

1) This is not a peer-reviewed research paper. This is commentary, maskerading as a peer-reviewed research paper. This article is not science, and the authors have explicitly chosen this, they don't claim it's science (yes, nature has a commentary section), or at least they don't claim it's even their intention to satisfy the standards demanded of papers in Nature. They're not even attempting that.

Presumably, that means this publication does not in fact satisfy the normal requirements to be published in Nature.

Nature, the (now many) magazines have a bunch of sections that only have to do with Nature, the company, and not with science or peer-review. The editorial, the commentary, the Q&A and I'm sure that's not all of it.

2) These people are sociologists. Sociology makes no claim to be politically neutral, and therefore the criticism that it's political is quite justified, even against the entire field. Of course, also doesn't mean it's wrong, but sociology is a science that, if you put it very directly, "isn't about being correct".

To be fair, both authors are academics boasting credentials outside of sociology. Medicine, economics, ... very impressive careers. Which is probably why this was published.

candiodari··on Paul Alexander, ‘the man in the iron lung’, has died
This is something people don't seem to realize. Often you hear people talk about how difficult it is to be born blind.

Yet everyone I've ever spoken to that happened to has a simple answer: "not difficult at all. It isn't there". One put it like "I'm also not a tiger and can't jump 8 meters high. It's similar".

Very strange how adaptable humans are sometimes.

candiodari··on Apple announces ability to download apps directly from websites in EU
I read that as: if Apple wants to allow installation of programs ("apps") on IOS, it must allow, free of charge, others to do the same. Free of charge to everyone. Free of charge to alternative app stores, free of charge to developers, free of charge to apple customers, ... free of charge to anyone. As I said, I'm no lawyer, but that is definitely a valid interpretation to me.

What exactly is unreasonable about that reading?

candiodari··on Apple announces ability to download apps directly from websites in EU
I think the original reading is pretty damn correct. It says apps should be able to access the platform "free of charge". Maybe I'm wrong but it seems to me that the reading that limits this to special API access is the tortured reading.

Besides, even Apple's reading is not what Apple is doing either. They're saying that ANY API access that is possible should be done free of charge. Ok. That INCLUDES app installation of course. It does not specify WHO doesn't get charged, which Apple then takes to mean those alternative app stores don't get charged, but the app owners do? Now THAT is tortured reading. Obviously that means NOBODY gets charged. Not the alternative app store, not the application being installed. Apple is not complying with their own reading either.

It seems to me pretty clear. Either interpretation, apps should be able to run on ios free of charge.

candiodari··on A Dutch Quandary Offers a Glimpse of a Deepening Problem for Europe
I believe it doesn't even need to go that far. EVERY party is running on a platform of "we're going to stop those evil assholes". Left says that about the right, the right says that about the left, greens, ... everyone.

We need a party that goes back to "we'll kill programs and then start a big program to do X". Better public transport. Housing in large cities. Land on the fucking moon again if you want. SOMETHING. With the power of government they're all very much solvable, but no parties are even interested.

candiodari··on Sweden Is a NATO Member
The problem with that is simple: if one side believes they are at war, 2 sides are at war.

And Russia has explained why they believe they are at war, the insane dictator at the top explained that in the 13th century it was really unfair that some army marched from Poland to what is now Chechnya (even Tucker Carlsson frowned as hard as he could and needed about 6 seconds to collect his jaw from the floor) and therefore Ukraine isn't a "real" nation and should be part of Russia. And that, in any case, Russia has the borders it chooses, as does any large country.

candiodari··on How Apple Sank About $1B a Year Into a Car It Never Built
And yet it shows, yet again, the same about companies (and management). Steve Jobs was lucky, right time, right idea, right people, massive interest in the right things in the market, right location, right hardware available, right access to market and stores, production capacity available (for reasons entirely unrelated to what Apple did) ...

Yes they worked, but they rolled a hard 6 not once, but 20 times.

In other words: what about Apple makes them able to repeat this in another market? Nothing. Not even Steve Jobs (and yes I know he's ... which is yet another reason ...)

candiodari··on Training LLMs from ground zero as a startup
Google python binaries are more akin to docker or even vm images, even if the actual technology used predates docker and even linux VMs. They contain something like a slimmed-down linux distribution, not just a binary.

EXTREME predictability (e.g. as never ever using the system's libssl), in trade for huge binaries. They go pretty damn far in this: you won't catch a Google binary even using most of libc.

candiodari··on Why do tree-based models still outperform deep learning on tabular data? (2022)
TLDR: Because tree-based models don't just outperform deep learning, they totally outclass deep learning on simple data.

But they don't scale with larger and more complex data. You cannot (realistically) make an LLM with XGBoost.

Kind of surprised how well Resnet and FT Transformer do though.

candiodari··on What the CEO wants you to know (2023)
Luckily in the supposed alternatives, like the middle east (which has lending+interest, they just lie about it for religious reasons. "It's not interest, it's just a administrative fee per dollar lent. Not Interest! No sir!"), or China. Or China's efforts in Africa, for example. Luckily there nature is doing well!

Oh wait a second ...

candiodari··on [dead]
This is a standard negotiating tactic, right? Ask for something really difficult AND for something entirely unacceptable. Then, after pressure is applied, drop the entirely unacceptable demand and "show your flexibility". And there you go, you made a concession, now the other party must make a concession ...
candiodari··on Google abandoned "don't be evil" – and Gemini is the result
Obviously once you go back 200 years 99.9% depictions have to be fake and once back 2000 or so years an outright 100%. And everything that we do have is also cherry picked and faked, by the original authors, but still fake. Pictures are cherry picked (and we don't have any going back even a few centuries) and paintings are outright fantasy.

Just look what a giant Napoleon is in all his paintings (or even the size of his grave)

candiodari··on Google CEO tells employees Gemini AI blunder 'unacceptable'
Why do you assume that real intelligence is not just heavily editorialized when it's in humans? In fact I'm pretty sure it is.

If you mean that neural networks are flawed things, not "perfect" intelligence like the Star Trek computer, then yeah, that's pretty obvious. Probably Star Trek assumed that perfect rule-based AIs, like IBM tried to build in the 1960s and 1970s were going to really work at some point, but keep having their "typical" fraud (such as getting totally stuck on a very regular basis for very dumb reasons)

candiodari··on Google CEO calls Gemini completely unacceptable, vows to make structural changes
> describing something as wrong is easier than saying what's right

This 1000x. This problem you also see with education of children. You have 2 problems:

1) there's a VERY large number of possible depictions. Let's say 10 million. Of those, a VERY small number are correct/acceptable. Let's say 10.

Then it's easy to see. A correct example has 1/10 = "10%" of all information so to speak. If 100% is what you need to be guaranteed a correct classification of all possible depictions.

Yet a negative example has 1/10,000,000= "0.000001%" of the information you need to classify all examples correctly. If you attempt to create correct behavior by giving wrong examples ... it'll be a long day.

A positive example, telling an AI (or a child/student) what to DO has ~10000 times more information than telling them what NOT to do, in this example. In practice, you'll find the difference is even more extreme.

2) BUT there is a problem, that's also always brought up with wikipedia. Giving positive examples requires that there are examples that everyone agrees with. And we all know there are rather serious problems with that. A positive example HAS to be in the intersection of what all groups find acceptable. And ... well easy examples are always wars: Who is right? Who started the war? Is it justified? Applied to Russia, Ukraine, Israel, India, Sudan, ... but in practice a lot of issues, some not at all that controversial (which word is correct "New York" english? Bupkis or bupkez?)

candiodari··on Google CEO calls Gemini completely unacceptable, vows to make structural changes
And yet this is exactly what management demands. I guess it's one aspect of the self-sabotage.
candiodari··on Mistral Remove "Committing to open models" from their website
But if you have cloud based models ... same with mainframes 30-40 years ago ... it doesn't matter. Because the power cloud gives to the model owners is 10x more than the strictest intellectual property laws do.

I don't believe, for example, that there's any intellectual property law that would let me yoink my intellectual property from you for any reason after you've bought and paid for it.

In other words: I think the capitalism discussion is kind of pointless here. Capitalism isn't what gives these companies power. It's the cloud. Mainframe computing 2.0.

candiodari··on Every single new Google product
> You're right. It's a very dynamic area, so more change could come at any time. For now, I'm glad they've reduce the overall number of names being used.

As for AI apps. There is "Assistant", "Gemini", "AI studio", "Gemma", "Natural language" (yes really, as in "Google Natural Language", not sure what it does), "Duet", "Duet for workspace", "Cloud Vision", ... This you call reducing names?

https://cloud.google.com/products/ai?hl=en

> The general consensus seems to be that for a long time Google employees were lazy/complacent/taking-it-easy. If it was as bad as has been said,

This would make sense ... except then I look at a bunch that I know were fired and I have to say, I don't agree these people were being lazy. Plus ... loads of people with exceptional skills, that found new jobs, in some cases even before the layoff announcement, and all but one I knew in less than 2 months. This, to me, does not sound like lazy people.

Besides, that wouldn't even matter. The problem with lazyness is that the choice is clear: what do you want? A very lazy expert? Or a very eager incompetent worker?

The truth there is simple: A lazy expert will perform between 0 and incredible. A very eager incompetent worker will perform between destroy the project entirely and slighly above 0.

And in my experience what works well for software is a mix. Lazy or not, medium to expert ... 80%, eager juniors 20% or so. And the juniors cannot be entirely junior (finished a decent education, capable of serious programming on day 1).

candiodari··on Institutions try to preserve the problem to which they are the solution
The problem is what San Francisco already experiences: it will attract more homeless from elsewhere for 2 reasons, one bad, one horrible

1) homeless will want this, and move by themselves

2) other state and city governments will dump homeless on San Francisco as a cheap way to "solve" the problem in their city.

candiodari··on Every single new Google product
You're assuming Google AI product rug-pulling, renaming, product replacing antics are now over. These included even a layoff of the assistant team, guaranteeing that everyone on even the current Gemini team is looking for other options, to avoid finding their badge doesn't work one morning.
candiodari··on Yishan Wong: "Google's Gemini issue is not about woke/DEI"
> Billions have been invested in this technology, and it sort of works!

And yet all reviews are like this, or (often) much worse: https://www.androidpolice.com/google-gemini-further-impressi...

candiodari··on The EU Wants Its Own DNS Resolver That Can Block 'Unlawful' Traffic,TorrentFreak
If you had the force of law behind you? Absolutely.
candiodari··on Yishan Wong: "Google's Gemini issue is not about woke/DEI"
Indeed. I've yet to see a single youtube evaluation video of Gemini that didn't at some point point out that Gemini is hallucinating or trying to convince the user of something that's total bullshit.

It's pretty clever about it. Doing this like. "Smart, clear, eloquent fact A, phrased longer than necessary. Bullshit (short sentence). Smart, clear, eloquent fact B, phrased longer than necessary".

But the bullshit is always ... pointing in the same direction. I must say I wonder if it isn't exactly the result of what is being blamed. The "wokeness". Regardless of the actual ideology people are trying to impose on the model ... at least in some cases the model is lying to make it's answers satisfy what is obviously a political ideology imposed on it.

Because look at those pictures. To a transformer, pictures are stories. A long sequence of tokens that convey thoughts. They are very largely correct. There's a clear Hollywood influence, as you'd expect, but otherwise they're mostly correct. If you took the total amount of information in those pictures, I bet you'd find 99% of it matches the training data! And making the characters black or asian is a little hallucination somewhere in the middle of the picture's story. And, surprise! Those modifications are there ... because that's EXACTLY the ideology Google is trying to impose on the model.

Could it be that the model is lying ... because it thinks that's exactly what you've asked it to do? "System prompts" mostly are just pasted before the input. If the system prompt says "assume all historical accomplishments were by black or asian people", then the model will assume that's what you want!

Makes you wonder ... how often are "hallucinations" the result of the model not making a mistake, but "purposefully" lying to make it's answer comply with a particular worldview? Because that's what you asked it to do, if you look at your full input, including Google's system prompt?

Hell, I'd ask the same question about a lot of human answers too. As soon as a subject becomes even a little bit controversial, people outright lie en masse. I know nobody wants to admit it but that's exactly how humans work.

candiodari··on AI is a 'Glorified Tape Recorder,' Says Theoretical Physicist Michio Kaku (2023)
cough https://www.google.com/search?client=firefox-b-d&q=What+are+... have you heard of "site:reddit.com"?

I will say that "site:reddit.com" is merely a poor man's way of telling Google "don't give me low quality machine generated results". Which, ironically, is exactly what you want ... from an AI!

Funny ... it's like a text based version of the plot of terminator: only machines ... can protect you from the machines!

I do think you're right btw, but for me the best applications of AI are, well, applied. As in "write tests for code X", "use library Y to ...". The code is definitely not useful as-is, far from it. But ...

candiodari··on AI is a 'Glorified Tape Recorder,' Says Theoretical Physicist Michio Kaku (2023)
Same can be said of my manager.
candiodari··on Jeff Dean: Trends in Machine Learning [video]
And all the people that actually did the inventing promptly left Google.

(yes I get there's 2 exceptions)

candiodari··on Why do so many big tech companies have awful apps and many indie apps are great?
> And if a miracle occurs, and when a huge organization produces some successful, well-scoped product, it is inevitably discovered that someone was given a budget and more or less ignored by the chain of command, and was able to ship something good under their noses—and if it is successful, expect version 2.0 to be have been meddled with.

Well of course. If everyone in power at that organization DOESN'T meddle with it, they're risking huge promotions to the new guy. To be bypassed, in the best case, or outright lose to them.

In huge companies, sabotaging others is an essential part of the way managers climb the ranks.

candiodari··on The EU Wants Its Own DNS Resolver That Can Block 'Unlawful' Traffic,TorrentFreak
You could just block DNS outbound from the continent entirely. How does that violate privacy?
candiodari··on The EU Wants Its Own DNS Resolver That Can Block 'Unlawful' Traffic,TorrentFreak
How do we get from DNS to Human rights? Why would they get involved?
candiodari··on Carbon Capture Facility in World Creates 25x More CO2 from Use of Product
You can joke about Trump, but he may get elected again. Which means this position won't change, at least for 9 months + let's say 1 year, potentially for 5 years, and we don't know where we'll be then.

> The majority of members advising the UN want it regulated, sure, and that's a sensible position due to unforeseen risks and the fossil fuel industry using a small amount of geoengineering that counters a portion of the problem as an excuse to double down on activity that continues to cause most of the problem.

Yeah, but read between the lines: they don't want it regulated, they want it outlawed, and, at least to a significant extent. They want it outlawed, firstly because if it isn't outlawed their power will disappear. And secondly, because no matter how you change the climate, it'll have the same problem any large infrastructure seems to have. Geoengineering WILL be incredibly good for some regions (who therefore may choose the lone wolf approach), AND will have devastating consequences for other regions. Someone will benefit greatly, someone will suffer.

And, frankly, if we're being honest ... "the West" (Europe/America/Russia/China/Northern Africa... even India) will benefit greatly, if we're honest. Africa and South America, and Indonesia will suffer.

But of course there's the identified problem. A decent-sized city could undo global warming (amplifying it is actually harder) BY ITSELF. Something like London or Paris or New York definitely could do it with the city's budget and land, and maybe even something like Milan would be big enough. Which brings the question: countries like Congo will greatly benefit from stopping global warming. They can, if they want to. What do we do then? Because, ironically, if they stop global warming through geoengineering, it will be a huge problem for "the West".

candiodari··on Carbon Capture Facility in World Creates 25x More CO2 from Use of Product
I think you will find the UN position on geo-engineering is quite simple:

1) It will work. Not just the technique suggested above, but a lot of of them.

2) That's bad: it will "normalise" dumping CO2 into the atmosphere and take away power from the climate organisations of the UN, including UNEP.

3) They want it outlawed. Why? Because geoengineering can be so effective that a single decent-size city can counteract all or a significant portion of all global warming.

4) They warn that this will produce large scale unintended consequences. Such as stopping water supplies that depend on glacial melting (and for the first X years (decades?) will actually almost entirely stop glacial melting, which "may" make places like California, the Indus valley and most of inner Mongolia, maybe all of inner China, unliveable. Also it will probably start expanding the sahara again (due to climate change it's shrinking now), which will also destroy communities)

(why? It is not possible to pump large amounts water either upwards, or along long distances. We don't have the power for that. So even if you collect the water desalinated, it can't be used to make those places liveable for anywhere the amounts of people currently living there)

Recent example of UN position of the matter: https://www.climatechangenews.com/2024/02/15/switzerland-pro...

← PreviousPage 6 of 34Next →