MrEldritch

For future reference, "Luboš Motl" and "puts it so viciously, dismisses without genuine inquiry or curiosity [....], and plays so much faster and looser with his ideas than the author he critiques—you can taste his contrition" are pretty much synonymous. The man's a troll with a PhD.

"Man why would they do that?"

Security reasons, obviously. Can't have those poor, gullible "users" copy-pasting malicious code into the developer console, now can we?

As someone who's worked at a grocery store - the 4-digit code is also printed on those little produce stickers, and will generally also be on the labels with the prices on the shelves. Use this to avoid searching by name!

True, but all the nonlinear optical effects I'm aware of only really start to matter at very high intensities - so wouldn't really be applicable to the kinds of scenarios they envision, like directly feeding it images seen from ambient light.

The short answer: Nonlinearity isn't just important for deep neural networks - nonlinearity is deep neural networks. Without a nonlinear element in between linear layers, the "deep" is meaningless - a "deep linear network" is precisely equivalent in power to a simple, one-layer linear classifier[1]. (Because if all you're doing is a bunch of linear transformations, you can't do anything you couldn't do with a single linear transformation.)

As far as I can tell, your understanding that this is just a linear function is precisely correct - which means it can't do anything that a simple linear classifier can't.

[1] I suspect that the reason this has multiple layers is because of the physical constraints of the system that prevent a single layer from being an arbitrary linear function of the inputs. The light from a specific pixel can only get effectively diffracted so far, so they need to cascade multiple layers to make sure that all the inputs can contribute to all the outputs. It still ends up being equivalent to a single linear transformation.

Nostalgebraist is the name of a Tumblr user, whose blog tagline is "Trees are harlequins, words are harlequins" (A quote from a Nabokov novel.)

This gets automatically appended to the page title for anything that gets posted on their Tumblr, hence why it's showing up here.

As for the rest, this is explaining a recently popular neural network architecture known as the Transformer, responsible for advances in natural language processing like GPT-2.

You are positing a catastrophe of such magnitude that "well, the value of my Libra has gone to shit" is the least of your worries. More appropriate concerns in such a scenario include "How can I acquire enough BDSM gear to wear and skulls with which to adorn my car to survive in the forthcoming Mad Max apocalypse."

One thing that frequently comes to my mind - rather than trying to convince people to move to a general-purpose programming language (and ignoring all the reasons that people turn to Excel in the first place), is it possible to turn an Excel-style spreadsheet into a general-purpose programming environment that is powerful enough to do that sort of stuff?

People really underrate Excel as a programming environment, honestly. Yes, it's crippled and leads people to produce massive gross un-debuggable hellsheets ... but there's reasons (beyond just "it was the only usable software that could be run on office computers") that end users with a problem to solve keep turning to it despite those flaws. The combination of reactive programming, data-first visibility, no hidden state, decent approximations to structured programming by way of click-drag-and-copy-paste, and being able to reference variables and values without needing to name them has some kind of magic to it.

There's quite a bit of interesting research I've seen from Microsoft on how to take something like Excel and turn it into a non-crippled programming environment - spreadsheet-defined functions (including recursion, lambdas, and higher-order functions natively in the spreadsheet environment, without having to drop into VBA!), dynamic arrays, an alternate computation-first textual view that exists simultaneously with the data-first spreadsheet view, first-class complex data structures ...

Between HyperCard, Jupyter/Mathematica notebooks, and Excel, there's a lot of common features that seem to point to a vision of a truly useful end-user-accessible programming tool, something that would let people write real, useful programs intuitively and quickly enough to be worth building them themselves for whatever they needed them for, without forcing them into condescendingly-simplified drag-and-drop "visual programming" codeless clunkfests.

I believe he acquired it through ex culo research, so him providing his source would be rather inadvisable on an SFW website.

I can't get this running at all, which is a shame, because I find the thing you're describing incredibly fascinating. Any advice?

The funny thing about all this is that I think - despite all his bluster - that Wolfram is basically right about how uniquely powerful this stuff is in terms of making advanced symbolic programming with complex data accessible and visible, and in providing a rich-text notation to express computations. I think that Mathematica genuinely captures a piece of what it would mean to have a computer as a mind-amplifier, a tool rather than a mere appliance, one which is frictionless and accessible enough to be usable by everyday people. I think he's actually on to something (as opposed to merely being on something).

...The problem is that it's attached to the rest of the Wolfram language - a painfully awkward Lisp-ish thingy dependent on an expensive closed-source platform and a standard library that seems to contain everything you could ever possibly need to do ... but which is so intimidatingly enormous that you can't keep track of it, which turns every program into a trek through the documentation in case there's something there that already does what you need (and then trying to figure out how to plumb all these bits together.) You can do virtually anything with the Mathematica standard library ... which is good, because trying to actually write de novo code is despair-inducing.

And Steve doesn't seem to be able to recognize that Paragraph 1 is ultimately crippled for widespread utility by Paragraph 2.

I do hope that eventually someone manages to take the notational insights of the Wolfram language and apply them to some other platform. There's a vague vision in my head of something that combines aspects and insights of Wolfram/Mathematica, HyperCard, Jupyter, and Excel[1] to create a truly flexible and accessible end-user programming environment.

[1] people really underrate Excel as a programming environment, honestly. Yes, it's crippled and leads people to produce massive gross un-debuggable hellsheets ... but there's reasons (beyond just "it was the only usable software that could be run on office computers") that end users with a problem to solve keep turning to it despite those flaws. The combination of reactive programming, data-first visibility, no hidden state, decent approximations to structured programming by way of click-drag-and-copy-paste, and being able to reference variables and values without needing to name them has some kind of magic to it. There's quite a bit of interesting research on how to take something like Excel and turn it into a non-crippled programming environment - spreadsheet-defined functions (including recursion, lambdas, and higher-order functions natively in the spreadsheet environment, without having to drop into VBA!), dynamic arrays, an alternate computation-first textual view that exists simultaneously with the data-first spreadsheet view, first-class complex data structures ...

I doubt this will happen. For one thing, the dominant targets for Linux are servers and embedded systems - both of which are cases where resource overhead is pretty important. Running a Windows to run Linux inside of incurs a pretty substantial overhead, so I really can't see any case where developers would be willing (or even necessarily able) to spare the resources to install Windows on their formerly-Linux-only targets just to maintain the convenience of getting to develop for "Linux" on their Windows PCs.

And desktop apps developed for WSL will simply never be Linux-compatible out of the box at this rate, given that WSL does not contain any graphical or UI functions; MSFT has been pretty clear about this right from the start, and so I really can't see a scenario where people mistakenly believe just developing user-facing applications for WSL will make them "just work" on Linux, given that all the UI code is on the Windows side.

"Downvote as disagreement" is such a pervasive pattern that I Find myself doing it automatically even when I know that's not what downvote is "supposed" to be for. I suspect it's because I'm more likely to view posts that I disagree with as being poorly thought-out.

This even extends to platforms where downvoting is not a thing - witness 4chan's struggles with "sage is not a downvote" ("sage" being a function which allowed you to reply to a thread without bumping it to the top and extending its lifespan - again, intended to mark a thread as low qualit) until they finally gave up and made saging invisible.

Given how thoroughly inescapable downvoting-to-disagree seems, I wonder if it might not be better to just declare defeat, allow downvotes and upvotes to mean nothing beyond a signal of protest or agreement, and have a separate "this comment is low quality / this comment is high quality" buttons.

I have a friend who believes that it is genuinely possible for a life to be net-negative expected utility, making suicide the right answer sometimes, and that he either is or may eventually be in such a scenario.

He is too afraid to talk to anyone about this except under an online pseudonym, because he is too afraid that if he actually talked to a psychiatrist about this, he would be immediately and involuntarily hospitalized for suicide risk.

He is terrified that he will soon not have even the option to commit suicide - even if it is truly better than continuing to face a life not worth living - because of precisely this kind of well-meaning ubiquitous monitoring and predictive warning. I'm really worried that he might just try and kill himself pre-emptively while he thinks he still has a chance.

Even sculpting and design apps would strongly benefit from force feedback. In fact, those are some of the specific things I was thinking of where the kind of precise movements that closed-loop control enables are critical!

When you're drawing, for instance, the pressure and friction of the pencil against the page is pretty critical for accurately feeling and controlling your movement. You cannot just wave your finger in the air with the same precision or control you can draw a line with a pencil.

Likewise for sculpting - your hands are far more sensitive to precise shape than your eyes are, and being able to feel the material resist against you lets you use precise amounts of pressure to judge how you are deforming it.

As long as VR inputs involve waving your hands in the air unsupported, they will never be able to achieve the same level of precision input and utility for serious tasks as a mouse or a drawing tablet.

The fact that we don't have good force-feedback technology yet is basically why we don't have great VR input devices. Being able to make fast, precise, confident movements depends on having resistance and collision to take the brain's motion control from open-loop to closed-loop. Even supposedly feedbackless touchscreens have the benefit of screen friction and a defined rigid plane.

And people have been trying with force-feedback; it's just really hard to make it work with something that allows for wide range of motion, isn't massively cumbersome, and costs less than a new car.

Why would you assume there's an explicit blacklist? I'd think it's just that people independently note who gets a lot of bad press.

Some people are just potentially a liability to provide service for because of the risk of mass activist backlash and terrible PR if someone decides you're supporting them and it gets traction online.

This same behavior is how certs usually work. Stuff with expired certs just does not run after the expiration date; that's because the cert tells you what server to ask for authentication, and if you have an old cert, there's no way to be sure that the original issuer is still the one in control of that domain.

That's what I meant, yes. (I didn't even know that KeePass could be integrated with the browser, I've just been manually copy-pasting)

There was also an issue last year where every single Oculus Rift was essentially bricked because they forgot to renew a cert (apparently, what with the chaos of the Rift launch and the Facebook acquisition between the cert issuance and expiration, they just kind of ... lost track).

It took like two days before there was any kind of fix available, and they couldn't even roll it out automatically because the expiration had also disabled the auto-updating.

Bugzilla issue here: https://bugzilla.mozilla.org/show_bug.cgi?id=1548973

It's a signing cert failure, not an intentional move. Still... this is going to destroy user trust in Firefox.

This is a goddamned disaster. I'm just thankful that I use an offline password manager, but even still ...

I like FF, don't get me wrong, but this is going to absolutely fucking destroy user trust in Mozilla. This kind of incompetence, on a browser scale, is breathtaking.

As a temporary fix, go to about:debugging, and click "load temporary addon", then paste in the download link of the missing add-on. Then just try and not restart Firefox until they fix the broken cert.

By "alleged patent timing", do you mean the patents are only alleged, or that the expiration dates of those patents are only alleged?

Because either one of those statements is ... confusing to me.

While that's a solid point, I think a critical part of why people look down on "clickbait" is the additional ingredient of "withholds key information to force readers to click ... and that key information makes the content substantially less interesting than the headline implied." The reason we call it clickbait is because it implies that it would clearly not be something you'd want to click, were the hook left unbaited.

Otherwise, it's not all that bad - just clever headline choice. (The point of headlines has always been to intrigue prospective readers, after all.)

Since this article is a substantial piece of journalism that remains interesting if you know the tl;dr, rather than an insubstantial chunk of #content, I don't think it really counts as clickbait.

It doesn't really matter whether you would or would not take the gamble; it depends on whether the FDA would be willing to let you.

(I'm not saying this as some kind of rrrgh-gubmint-bad thing. It's pretty normal for the government to have authority over what gambles are acceptable to offer in what circumstances, and which ones are exploitative and detrimental to even offer; this is the logic behind, for instance, regulations of actual gambling.)

It'll probably end up "solved" in a similar way to email span being "solved" - if global spam starts to become a problem, you'll have big blocklists of spammy instances and general suspicion towards federating with any new instance that doesn't look sufficiently trustworthy.

Honestly, I think this is the wrong way of looking at things. The Fediverse isn't really like a single social network that's shattered amongst many instances ... it's more like a bunch of individual, separate hangouts, in the vein of classic forums and IRC chatrooms and fandom websites, which have the ability to talk to each other and share stuff if they want.

It's not so much about "blocking" mastodon.social as not choosing to link up with it - because your instance is a specific community, and actual human social relations thrive when they can choose who they're hanging out with. So you choose to talk to other smaller, more diverse communities that you can actually get to know a good chunk of the people in them, instead of just drinking straight from the firehose that is a big instance.

Federating with another instance is basically two groups of people deciding to hang out together, effectively merging. Federating with mastodon.social means, effectively, merging your community with mastodon.social ... which means submerging it beneath the mass of mastodon.social's much greater activity.

A vibrant Mastodon isn't a more-obtuse recreation of the Twitter experience. Twitter is already Twitter. It's about fostering all the things that Twitter isn't - personal and Dunbar-scaled and locally moderated and diverse.

The thing about 4chan is that it's got a shitton of boards, and those boards are like kind of like, say, subreddits - they're pretty self-contained and have distinct communities. So, say, you have places like /pol/, but you also have places like /tg/, the "traditional games" forum (i.e. Magic enthusiasts and RPG players) which turned into a bubbling hub of creativity and collaborative storytelling. (and shitposting.) 4chan was abrasive everywhere, but it's hardly vile everywhere.

The experience I found on 4chan is pretty unique, and I don't know how well I can describe it, but I'm really glad that there was somewhere like it.

That said, 4chan's culture definitely started to decline around 2014~2015ish, around the time of Gamergate and the Trump campaign. Whenever 4chan's "hub of vileness" reputation starts getting really public in the news, the result is that a bunch of new people who think "hey, that sounds like exactly my jam!" start flooding in. And those two events both drew a huge amount of attention to 4chan, and not the kind of attention that brings good people. (It didn't help that a huge chunk of the old userbase fled around the time of Gamergate due to controversies about it on the site itself.) /pol/ was always a little leaky, but it started to get a lot worse around that time. Not to mention that /tg/ itself was already on the decline because of the positive reputation it had gained elsewhere on the Internet as various stories originally found on it were posted elsewhere, and so it increasingly became full of people who wanted to hear those kinds of stories but weren't actually RPG players themselves. There's a lot less positive to recommend the site these days, and a lot more negative.

It's not particularly baffling if you think that Western nations are taking these stances for approximately the same reasons as Russia and China are.