Posts by BinkyTheHorse 105 publicly visible posts • joined 23 Feb 2013
Oh, they probably already can. This is a smoke screen, a way to legitimize the output of their generative models, in the face of the training data for such models becoming more and more scrutinized for copyright breaches. So, more like:
"See, see, this one person gave us the go-ahead on this output! Never mind the mounds of copyrighted data used to train the model, and absolutely necessary to produce this output! "
So, you can take a neonatal human brain, connect its visual cortex to a carousel of Cubist images, and get new Cubist paintings through the motor output, is what you're saying?
The point here is that learning style in humans is more involved than just ingesting images. Humans train to know what an artistic style is, for starters. In end effect, the process is considerably more divorced from any "source" images than what training an ANN looks like. The real issue at hand is what is, and what isn't, a derivative work. With humans, there is, when done professionally, less of an evidently demonstrable connection between any applicable, copyrighted source, and the output.
I also despise how dualism insidiously comes of the woodwork in this sort of discussions.
However, I'm afraid that this is essentially a request to prove a negative, with extra steps. We don't know what exactly shapes cognition in human brains, and especially not that it can be reduced to a sequence of matrix operations, like with ANNs.
> But style isn't, and shouldn't, be copyrightable.
Yes, and as soon as the models just use style as training data, then that argument becomes valid.
But, right now, the models in question are trained on actual, copyrighted images.
To preempt the "oh, but humans do the same" – prove it. Demonstrate how style is learned by a human being purely from an image (or a set of such). Show the exact sequence of biological process that ends up with that – similarly how the exact sequence of operations can be shown for an ML model. Otherwise, it's all just handwaving.
Not a lawyer either, but pretty sure decriminalization means just that i.e., legislating something that was a crime to no longer be a crime.
Law enforcement officers selectively turning a blind eye, and calling it "decriminalization", is a perversion of the type of legal system we have in most democracies. Law enforcement should be limited to, well, law enforcement.
At worst, initiatives like this might be cynical ploys to pump up statistics later on, by getting drug users to get their guard down around police, and then throwing the net at a convenient point in time, later down the line.
Whether the output is "art" or not is completely insubstantial. The issue is whether it constitutes derivative work. Applies to other stuff, like, you know, programing code for example.
The entire debacle is clearly seen as an opportunity to use "AI" as a magical IP-clearing black box by the corps.
Icon for the treatment professionals relying on copyrightable work for their livelihood should tolerate, as probably imagined by the OP - who apparently prioritizes Schadenfreude over not living in a corporate dystopia.
> [...] Now when you try to install Firefox on the current Ubuntu, even when specifically using the .deb package, it silently replaces it with the snap version. [...]
The most prevalent story behind that is that this was being pushed by Firefox maintainers, not Canonical. And the replacement PPA is run by a "team" that includes Canonical engineers, so that further deflates the Canonical == Great Satan hypothesis.
All the features of actual Markdown, paraphrased from the original doc: Headers, Blockquotes, Lists, inline and Block Code (w/o syntax highlighting, references, etc.), Horizontal Rules, Links, (two kinds of) Emphasis, Images (w/o captions).
For anything else – and I suspect that in journalism you'd at least need stuff like tables, the occasional footnote, image caption, section link, etc. – you have to use some extension, or give up and write HTML.
So, to completely butcher that analogy: instead of a "sledgehammer", you now have a feather, with a set of pliers, a center punch, and a chisel glued to it, to break those roasted peanuts.
That's the advantage of Asciidoc in this context: instead of a set of pidgins, it offers a single, consistent syntax and semantics. All of this is also an enabler for some definitive documentation – and being able to look up 99% of general-usage features on a single reference page is quite a time-saver.
Well, wild fowl – and most other game – have a convenient "handle" that the robots usually lack (or if they do, they're probably not a great feel on the gums) :).
But if you're saying those were "gadget" robots that only weighted around 3kg, not "full scale" ones, than I wish to retract my statement, and apologize for the doubt. A hunting dog such as a standard poodle would be fully capable carrying it then, even if elderly.
As a bonus, this also explains why you had *four* of the bloody things.
OK, jake, you've had a good run over the last couple of years, but I have to call shenanigans on this one.
A standard poodle can easily drag along something the size of a robovac, so one and two are plausible, but three and four would require the pooch physically carrying both of them like sticks - even if your *elderly* poodle has that sort of determination and jaw strength, you could identify him by the bite indentations alone!
Also, cats not being interested at all at a shorter-than-them critter scurrying along in their territory, not even checking if it's edible? Yeah, right.
In conclusion, fun to read, but dial down the tall tale coefficient a little, or at least wait until Friday evening before reposting unmodified older stories such as this one ;).
"[...] usually you have thousands of parameters that you have to train."
No, "thousands" is for toy problems. Tens of millions is completely normal for real-world problems, with state-of-the-art nets being some orders of magnitude more complex still (GPT-3 has supposedly 175 billion).
Perhaps the researcher was confusing the number of "parameters" with "possible combinations of hiperparameters", but that's problem-specific, so doesn't map to ANN scales directly. Either way, a bit worrying to potentially see the relevant researchers being under such a misconception – hopefully it was merely a mental hiccup.
Regardless, progress in this problem is welcome in any case. Icon related, especially in the Summer.
The Dunning Kruger effect in this post is off the charts. You do realize machine learning is significantly older than deep learning? Or that most "machine learning algos" have a degree of interpretability that makes neural networks an outlier in this regard? Or that significant work has been put into making deep learning more debuggable and interpretable (the "dog in snow" paper is 4 years old FFS)?
This is essentially like wanting to scrap the entirety of software development after experiencing the multitude of shitty mobile apps. Don't blame the tool for the tools that use it wrong.
Oh, and another problem with this specific discussion is that it married two unrelated subjects - "is enterprise $ANYTHING a load of PR/marketing hogwash" (it very often is, but that's the fault of PR/marketing/sales, not $ANYTHING) and "is AI actually mature enough to be impactful at your typical mid-sized company" (a topic actually worth discussing).
So, El Reg, for the next time, please - in the words of a certain secret agent - phrasing!
Icon is me being prepared for the votes underneath the two posts.
Here's a refutation: that's an approximation of the definition of symbolic and subsymbolic systems.There are plenty of ML schemes that are fully capable of explaining themselves, most famously expert systems. On the flipside, one conclusion from the stated definition is that virtually every person is at least intermittently not "intelligent".
<rant>And this is the weakness of at least the "public comment" side of such "debates" - people with (next to) no grasp of the subject reinventing the wheel with their late-pub-night grade hypotheses instead of spending the 5 minutes it would take them to actually improve their relevant personal knowledge. It is evident even in this very thread (no, Elledan, that's not how working on improving a CNN model works, unless done by a complete amateur that spent the last 5 years under a rock). In the end, the subject of the discussion devolved into "hurr durr, dem computers r dum cuz i says so".</rant>
Sorry for the abrasiveness there, but there's only so many times one hears something in the key of "heheh, it's just a a bunch ifs" before they inevitably blow their top - essentially the equivalent of "they just sit in front of their computers all they and get paid for it!" for ML specialists.
[...] so not judge the researchers (in terms of their career development) on where they publish their research, but on the research itself.
But the current system is built on that. For reputable institutions, academics are evaluated on the performance metrics of the journals they publish in, the journals tend to only select sufficient-quality papers, and in turn those papers (ideally) reinforce the journals' performance rankings.
So for anyone suggesting the current (definitely imperfect) system to be scrapped, the odium is on them to provide a superior alternative.
This is a megapatch not for "Linux" (i.e. the entire ecosystem), but for the Linux kernel.
As another commentard already noted below, there are non-kernel solutions for handling NTFS writes which already work quite well, and have been for many years.
Why should there be a "call to arms" if the prospective contributor has put (charitably assuming) minimal effort into preparing the contribution to be usable, and better alternatives already exist?
(also, I'm pretty sure there's kernel-level support for ZFS nowadays)
This is a difficult issue, but there's at least one thing that can guide your understanding if you are, too, an immigrant.
Ask yourself: "What perceived problems, according to myself, do I see that the members of my nationality/ethnicity are bringing into the country I immigrated to?"
If you don't have as clear and immediate of an answer as you had with the examples of the "others", your current mindset will not lead to a good solution.
To recapitulate:
a) any existing observatories should be abandoned and millions spent on moving that infrastructure into orbit - because, as we all know, science is not criminally undervalued in modern times anywhere in the world and always receives adequate funding;
b) all scientists should now rely on His Muskyness and other US-based launcher companies, regardless of not only their budgets, but availability, subject to trade and technology embargoes (maybe this is news to you, but not all people in countries the US has embargoes on are evil) ;
c) similarly, enthusiasts should either crowdsource money for their own space telescopes or just live with Elon's space grafitti. Because there is no value in the sense of wonder produced by tangibly observing space at one's convenience, and definitely not in inspiring the next generation of scientists.
Techbroism is rare here on El Reg forums, but when it shows, oh boy how does it show.
You do realize this Team America: World Police crap is exactly one of the reasons why the US patent system is so broken?
In modern times, it is difficult for a lot of businesses, even small ones, not to be obliged to interface with US law in some way. Hell, in IT, it is virtually impossible in some sectors not to rely on some US-based intermediary (credit card companies, the leading mobile app stores, Paypal, Steam, Patreon, etc.). The patent trolls wised up to that - it's much easier to target some small business in a foreign country than a US-based company with billions of VC funding that can potentially very effectively push back. Essentially, this creates a larger hunting ground with more defenseless prey.
So, maybe, instead of futilely frothing at the mouth on an Internet forum, you should be spending that energy on writing to the Congresspeople representing you. Because not everyone here can do that - but you can.
So if I someone would come to you and said that an apparition of Virgin Mary spouting fire out of her ass was present in their toilet from 03:15 to 03:30 AM, and they had proof they were indeed at home, and home alone, would you automatically believe them?
Careful with that line of reasoning. Just because there is a sole source doesn't automatically make it immune to critique. What if they cited maximum instead of average rates, would you use the exact same reasoning as well?
And unless that particular law is fucked up (which wouldn't be a surprise), you would be laughed out of court to argue for a "criminal party" where none exists - not according to any relevant judiciary.
Being arrested is not a crime in itself, nor does it prove criminal intent. So most of the clause proposed by the OP (apart from the jail time and actual crime-related legal sanctions) appears to be sound.
> For example, the EU ruling that the UK can't expel EU citizens who commit heinous crimes such as rape and child molestation.
So you'd rather they potentially walk off scot-free, instead of getting a proper trial, conviction, and sentencing?
> Or ruling that certain groups can be considered protected minorities even though their customs and practices are considered backwards and barbaric in the UK (No, not the obvious one, think again).
"Barbaric" is relative - case in point: milk in tea. Or, for a less extreme example, blanket state surveillance.
BTW, did you mean to imply that EU somehow forces the UK to render claustration, forced marriages etc. to be legal?
I'm not sure we played the same games.
While the dialogs are definitely "wordy" at times (but that's by design), they do influence what paths of solving problems you follow, what companions you get, what companions you keep, the availability of any unique items or abilities (yours or the companions'), how you interact with various factions, the possibility of secondary characters assisting you in some missions, and - not spoiling to much - even the possible endings. Certain choices can often take you out of a difficult fight - or into one.
Granted, this was more fleshed out in the second and third games (Returns is by comparison rather short and simple). Also, all three of them eventually got free "Directors Cut" versions, which are much more polished.
Ultimately, however, this comes back to what are your expectations for such games. If you're of a mindset of "clicking through" rather than "experiencing the setting" then, well, no wonder you didn't find it to your taste. And that's of course OK, but that doesn't mean certain design choices are bad.
PS. No, I'm not funded by Harebrained Schemes, I just like the stuff that they make ;).
[...] Amazon is a service developed in the US. Even when you buy something in the UK and it is shipped from a UK warehouse, you are using a US service. [...]
And if you buy French cheese, the process could be construed a trivial service, where, for an X one-time payment, you get M days worth of cheese.
That cheese provisioning service is undoubtedly "developed in France", correct? So, by that logic, imposing tariffs for its import is "too simplistic and doesn't work"?
One could even make the same argument about "splitting the company" w.r.t. shipping.
Be nice if they'd printed the list of 12 states so we know where the last bastions of sanity are.....
1. They're listed in the bloody linked article.
2. You do realize the main push for this was the scare against cell-cultured meat? Calling plant-based protein "meat" is already banned in most sane jurisdictions, so I dare say your criterion actually implies the opposite.
Icon for something I occasionally wish to inflict on those too lazy to read with comprehension, especially before spewing the consequences of that on various comment sections.
Or, you know, it could be that there are so many problems with Boeing's planes recently that people started associating all avionics issues with Boeing. But no, it must be because those dirty Europeons are disparaging Great American Enterprise™. Never mind that 2 planes crashed and hundreds of people died.
One day we may have discussions on such topics without the flag waving, but - very clearly - it is not this day.
I understand what you meant to convey, but it is physically impossible for the EU to have "started it", because the US has been playing the trade war game for waaayyy longer than the EU has existed. For you, as an Anglo-Saxon (or as a [possibly unwitting] apologist of that geo-cultural establishment) it may be an academic difference - for me, it's just another label for the same lube dispenser.
Re your 2nd paragraph - fair is fine, but keep in mind the negotiations hang on a very specific case that affects both the EU *and the EEA*. Put yourself in the shoes of the Norwegians - would you want a leaky border of subpar goods because the EU negotiators didn't have the gonads to make a stand where common sense dictates?
[...] You're all crazy. Just.. Crazy. [...]
Yes, because blatant protectionism such as very loudly proclaiming trade tariffs, banning foreign suppliers on spurious grounds, and - most relevantly - considering only US-made prior art as "prior art" (which El Reg covered a while ago) is not crazy. Preparing contingencies for such actions of a critical trade partner is, however, crazy.
Icon for how I see people that blindly believe in some markets being inherently "free", while others are not, despite any actual regulations present.
> Here is one of many studies you will find - https://www.scientificamerican.com/article/new-studies-link-cell-phone-radiation-with-cancer/
If you'd actually read that article, you'd find that:
- this is actually about two studies,
- the rats in the study, in most of the groups, were exposed to much higher radiation than legally permitted for RF (sounds familar to the infamous aspartame study, doesn't it?),
- the results are considered inconclusive by some quite important experts,
- the article specifically mentions that other studies contradict these ones in conclusions.
Not being able to look at the studies, it also looks as if the researchers were doing some statistical shotgun shooting, the problems with which are described in this XKCD comic.
You're confusing photon energy and emission power. If you emit a signal with a given power, it does not become more intense per area because the photon energy/frequency is larger. 1W/kg stays 1W/kg, otherwise we'd be magically getting increased total energy for "free".
The frequency does define harmful effects for given power, but the effect is not linear. In fact, higher-frequency microwave e.g. achieves less body penetration than lower-frequency. And, as Loyal Commander noted, before you get to ionizing radiation you still have visible light along the way.
Sorry for the nitpick, however... Please, be very wary of conflating a fallacious argument and an overtly similar, valid one. A slippery slope fallacy is an invalid argument, because it's based on an unsubstantiated implicit premise of "a slippery slope phenomenon exists here".
If you actually substantiate that premise, than it's no longer a fallacy, and using the shorthand "slippery slope" as a descriptor is IMO confusing at best, and detrimental at worst. It's essentially the same situation as "Professor Z will have his lecture at time T, because he said so yesterday" vs argument from authority, or "no true Scotsman can breathe in space" vs you guessed it*.
* Here's to hoping there's no canon continuity where Bruce Wayne dons tartan...
...shame about the circumstances :/.
Since that wasn't elaborated on in the text, the keyboard mod was called Livermorium. As you can see, it was a horizontal slide-out physical keyboard, which, AFAIK, is an object of great desire of many commentards in these trying phablet times.
Sadly, supposedly due to fabber shenanigans and lack of support from Moto, the project was ultimately cancelled, with only a part of the original backer order shipped.
The team behind it went on to working on a dedicated slide-out modern Android phone. Backers for Livermorium who haven't received their unit had the option to either get a full refund or get credit towards the new phone.
Overall, somewhat of a sad story, and it would be interesting to hear how it went from the Moto side of things.
Adding to DougS' post, take a look at the top-right corner of the RFC in question. Do you really think that Google would propose such a thing if it was going to adversely impact them in the long run?
Monopolies are especially dangerous because a) they arise naturally in an unregulated market, and b) a monopolist will eventually be compelled to try everything to remain one, while c) every other actor on the market suffers, even if the monopolist's actions don't appear detrimental at first.
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
