Hacker News Reader

> I also don't doubt that "every tech nerd's reaction is 'bullshit'", but only because supremely confident proclamations of universal truth that are soon proven wrong is pretty much the defining trait among that community (c. f. various proclamations that solar power is useless, CSI-style super-resolution image enhancement is "impossible" because "the information was lost", "512k should be enough...", "less space than a Nomad..", and everything Paul Graham has said, ever).

Notwithstanding the fact that you do have a point here, polemically phrased as it may somewhat ironically be, I just want to point out that the Paul Graham reference is probably not the best example of the "tech nerd community" trait you're describing. At least this particular community doesn't quite believe that everything Paul Graham say is true; a couple of examples:

* "Let the Other 95% of Great Programmers In": https://news.ycombinator.com/item?id=8799572

* "The Refragmentation": https://news.ycombinator.com/item?id=10826836

I could share a lot more HN discussions, and some of his essays where he pretty much describe the trait you're taking issue with here -- but I'm already dangerously close to inadvertently becoming an example of a tech nerd that believe "everything Paul Graham has said, ever", is absolutely true ;) I don't, and I know for a fact that he doesn't think so either (there's an essay about that too).

Grandparent doesn't understand information theory. True superresolution is impossible. ML hallucination is a guess, not actual information recovery. Recovering the information from nowhere breaks the First Law of Thermodynamics. If grandparent can do it, he/she will be immediately awarded the Shannon Award, the Turing Award, and the Nobel Prize for Physics.

True superresolution is impossible, but a heck of a lot of resolution is hidden in video, without resorting to guesses and hallucination.

Tiny camera shakes on a static scene gives away more information about that scene, it's effectively multisampling of the static scene. (If I have to hunch it, any regular video can "easily" be upscaled 50% without resorting to interpolation or hallucination.)

Our wetware image processing does the same - look at a movie shot at the regular 24fps where people walk around. Their faces look normal. But pause any given frame, and it's likely a blur. (But our wetware image processing likely does hallucination too, so it's maybe not a fair comparison.)

Temporal interpolation is still interpolation.

It's not temporal interpolation. It's using data from different frame(s) to fill in the current frame. It's not interpolation at all. It's using a different accurate data source to augment another.

Are you objecting to the premise that the scene is static?

Wait, why thermodynamics? The first law of thermodynamics is about conservation of energy not of information.

Same thing.

Super resolution can and does work in some circumstances.

By introducing a new source of information (the memory of the NN) it can reconstruct things it has seen before, and generalise this to new data.

In some cases this means hallucinations, true. But in other times (eg text where the NN has seen the font) it is reconstructing what that font is from memory.

But the thing is, in that case the information contained in the images was actually much less than what we are meant to make believe.

So if we are reconstructing letters from a known font we essentially are extracting 8 bits of information from the image. I'm pretty certain that if you distort the image to an SNR equivalent of below 8 bits you will not be able to extract the information.

Recalling memory of something different and using it to interpolate is a form of hallucination.

Wait until panta discovers image compression…

Lossy image compression creates artifacts, which are in a way of form of falsely reconstructed information - information which wasn't there in the original image. Lossless compression algorithms work by reducing redundancy, but don't create information where it wasn't there (thus being very different from super-resolution algorithms).

Not if it’s written text and you are selecting between 26 different letters. It’s a probabilistic reconstruction, but that’s very different to a hallucination.

> CSI-style super-resolution image enhancement is "impossible"

If it ever started to be used in court, Ryan Gosling will have a lot of explaining to do. I hope that day never comes.

You're both right, but they're more right because the subtle difference you mention is the problem they're highlighting: Apple went out of their way to be unclear and create subtle differences in interpretation that would be favorable to Apple, as a company should.

After the odd graphs/numbers from the event, I was worried it was going to be an awful ~2 year period of jumping to conclusions based on:

- Geekbench scores

- "It's slow because Rosetta"

- HW comparisons that compare against ancient hardware because "[the more powerful PC equivalent] uses _too_ much power" implying that "[M1] against 4 year old HW is just the right amount of power", erasing the tradeoff between powerfulness and power consumption

The people claiming this blows Intel/AMD out of the water need to have stronger evidence than comparing against parts launched years ago for budget consumers, then waving away any other alternative based on power consumption.[1]

Trading off power for power consumption is an inherent property of chip design, refusing to consider other chipsets because they have a different set of tradeoffs mean you're talking about power consumption alone, not the chip design.

[1] n.b. this is irrational because the 4 year old part is likely to use more power than the current part. So, why is it more valid to compare against the 4 year old part? Following this logic, we need to find a low power GPU, not throw out the current part and bring in a 4 year old budget part.

I really would suggest reading the Anandtech article linked above - I think it will help to clarify where the M1 stands against the competition.

Hate to draw a sweeping conclusion without any facts like this without elaborating, but, late for dinner :(: it's an absolute _nightmare_ of an article, leaving me little hope we'll avoid a year or two of bickering on this.

IMHO it's much more likely UI people will remember it kinda sucks to have 20 hour battery life with a bunch of touch apps than that we'll clear up the gish-gallop of Apple's charts and publications rushing to provide an intellectual basis for them, without having _any_ access to the chip under discussion, so they substitute iPhone/iPad chips that can completely trade off thermal concerns for long enough for a subset of benchmarks to run, making it look like the the powerfulness/power consumption tradeoff doesn't exist, though its was formerly "basic" to me.

That is not what Apple said. Quote:

"Faster than 98% of PC laptops"

[1]: https://d1abomko0vm8t1.cloudfront.net/article/images/800x800...

My quote was from Apple's MacbookAir page[1], including the footnote.

Your quote is just not very specific. On its face, it could mean every PC Laptop ever produced. I'm somewhat certain that every Apple computing device ever has beaten that standard. Even Airpods and the Macbook chargers might be getting close these days.

[1]: https://www.apple.com/macbook-air/

> Even Airpods and the Macbook chargers might be getting close these days.

I simply can't agree with these logical gymnastics.

2019 laptop sales are 9.3% of laptop sales for years 2010-2019 (https://www.statista.com/statistics/272595/global-shipments-...). The phrase "Faster than 98% of PC laptops" in itself is very general, an it's fair to assume that it's e.g. all laptops currently in use, or ever made - in part since this is a sales pitch, rather than a technical specification item. If we add 1990-2009 laptops to the statistic above, the share of purportedly modern laptops will just shrink, and substantially at that.

Then you have to add on top of that the consideration of what are the laptops people are actually buying and using. I can assure you that neither enterprise clients nor everyone outside of select, relatively spoiled, group people concentrated in just a few areas are regularly buying top of the line workstation or gaming laptops.

The parent is very much right, amongst techies there's an unhealthy tendency for self-righteous adjudication of the narrative or context.