Mark of the Many

[markofthemany]

What to Consider Before Trading Your Health Data for Cash

Don’t trade away your health data without considering the potential issues first.

[danshive]

In science fiction, AIs tend to malfunction due to some technicality of logic, such as that business with the laws of robotics and an AI reaching a dramatic, ironic conclusion.

Content regulation algorithms tell me that sci-fi authors are overly generous in these depictions.

“Why did cop bot arrest that nice elderly woman?”

“It insists she’s the mafia.”

“It thinks she’s in the mafia?”

“No. It thinks she’s an entire crime family. It filled out paperwork for multiple separate arrests after bringing her in.”

[dexer-von-dexer]

I have to comment on this because this is touching on something I see a lot of people (including Tumblr staff and everyone else who uses these kind of deep learning systems willy-nilly like this) don’t quite get: “Deep Reinforcement Learning” AI like these engage with reality in a fundamentally different way from humans. I see some people testing the algorithm and seeing where the “line” is, wondering whether it looks for things like color gradients, skin tone pixels, certain shapes, curves, or what have you. All of these attempts to understand the algorithm fail because there is nothing to understand. There is no line, because there is no logic. You will never be able to pin down the “criteria” the algorithm uses to identify content, because the algorithm does not use logic at all to identify anything, only raw statistical correlations on top of statistical correlations on top of statistical correlations. There is no thought, no analysis, no reasoning. It does all its tasks through sheer unconscious intuition. The neural network is a shambling sleepwalker. It is madness incarnate. It knows nothing of human concepts like reason. It will think granny is the mafia.

This is why a lot of people say AI are so dangerous. Not because they will one day wake up and be conscious and overthrow humanity, but that they (or at least this type of AI) are not and never will be conscious, and yet we’re relying on them to do things that require such human characteristics as logic and any sort of thought process whatsoever. Humans have a really bad tendency to anthropomorphize, and we’d like to think the AI is “making decisions” or “thinking,” but the truth is that what it’s doing is fundamentally different from either of those things. What we see as, say, a field of grass, a neural network may see as a bus stop. Not because there is actually a bus stop there, or that anything in the photo resembles a bus stop according to our understanding, but because the exact right pixels in the photo were shaded in the exact right way so that they just so happened to be statistically correlated with the arbitrary functions it created when it was repeatedly exposed to pictures of bus stops over and over. It doesn’t know what grass is, what a bus stop is, but it sure as hell will say with 99.999% certainty that one is in fact the other, for reasons you can’t understand, and will drive your automated bus off the road and into a ditch because of this undetectable statistical overlap. Because a few pixels were off in just the right way in just the right places and it got really, really confused for a second.

There, I even caught myself using the word “confused” to describe it. That’s not right, because “confused” is a human word. What’s happening with the AI is something we don’t have the language to describe.

Anyway what’s more, this sort of trickery can be mimicked. A human wouldn’t be able to figure it out, but another neural network can easily guess the statistical filters it uses to identify things and figure out how to alter images with some white noise in exactly the right way to make the algorithm think it’s actually something else. It’ll still look like the original image, just with some pixelated artifacts, but the algorithm will see it as something completely different. This is what’s known as a “single pixel attack.” I am fairly confident porn bot creators might end up cracking the content flagging algorithm and start putting up some weirdly pixelated porn anyway, and all of this will be in vain. All because Tumblr staff decided to rely on content moderation via slot machine.

TL;DR bots are illogical because they’re actually unknowable eldritch horrors made of spreadsheets and we don’t know how to stop them or how they got here, send help

[rock-cake-with-a-pin-in-it]

This stuff is cool and much more interesting than the general-AI doomsaying anyway (which I will drag in the tags anyway). :)

Here’s an article about adversarial attacks on image recognition neural networks, and here’s another one about how your training data may mean that your system learns the wrong thing, like “this photo has sheep in” actually being “this photo has places that sheep graze in”.

[markofthemany]

I was thinking of writing something myself about the trappings and limitations of current AI models but @dexer-von-dexer​‘s bit is so on point I’ll not rebuild the wheel.

The only thing I’d like to add is that the way those models work, there is inherent error. So called AI are currently tools that are good for slogging through monumental lists of redundant tasks that would be unapproachable without automation. The kind of tasks where it’s better to get some of it wrong than not do it at all.

The highest accuracy rates for AI operation involve very narrow use cases with easily machine defined tasks for example ‘this letter, not that letter’  ‘this clause not, that clause’ ‘this color, not that color’. As soon as we get into complex tasks like scenario and concept recognition we see a dip to the 92-95% efficacy range. Which is worth noting is a level of accuracy absolutely worthy of many applications, but is far too low to be acceptable for some.

Imho, the problems tumblr has been facing with the exploitation of minors and the porn-bot infestation absolutely warrant the use of AI. I’m not sure I would have set it loose with the same parameters, but this is the right kind of tool for the job. Without restarting the platform there’s nothing else that can work on the scale of porn-bot infestation like an AI algorithm. I can also appreciate that identifying legal vs illegal porn is difficult enough for humans that it’s unrealistic to try and guide coded associations for it.

All that said, even if their flagging-bot has 95% accuracy, 5% of all of tumblr is still a huge number of posts and is bound to contain all manner of ridiculous content that we as humans can recognize should have been categorized differently. Comparing inappropriately flagged and unflagged posts cannot demonstrate bias, it’s just inevitable availability of anecdotal examples with this size and complexity of task.

This is why flagging and sending to humans for verification is the accepted use model. Effectively sorting out the things that the machine can easily tell are blatant violations allows humans to focus on the tough cases. Even those tactics tend to have a margin of error, we’re only human after all.

[starwarsgif]

[markofthemany]

I wanna see jedi Leia.

[theconcealedweapon]

Anyone else think something is fucked up about the fact that of a group of 12 randomly selected women, all 12 have been assaulted by men?

[syrenaquills]

THIS IS SO WRONG

[redplaygrounds]

I wish I could articulate how I feel about this… I don’t even know where to start! OP was right on with their point, but also:

I get why people who have been assaulted would be biased, but isn’t everyone biased? and by trying to eliminate bias, all that I see achieved is that a jury is made up that doesn’t actually represent (proportionately) who the accused’s ‘peers’ are - because one in 4 of the female ‘peers’, ish, would have experience sexual assault. So if half of the peers are women, 6 women, then at least 1 of them should, to be representative, have experienced sexual assault. Which probably wouldn’t swing the vote, although it could. 

(That whole paragraph could be irrelevant depending on what you think ‘peers’ should mean but I think representative is ‘fair’, so rather than trying to eliminate bias, which is impossible, we would instead be trying to recreate the same biases of society. Which might not help actually, now I think about it!!)

Also WHY THE FUCK would the juror ask them all to describe it? Like, why do they need to know the details? To determine more specifically any level of bias the women could have? Maybe. But also WHY THE FUCK is it deemed ok to ask? 12x the trauma being relived plus 12x the triggering of others. 

I don’t know what country this was in - I’m assuming the US because so much of Tumblr and other social media is centred on US issues (and issues of what I would call white - Western - countries but that’s a rant for another day) and I don’t know what variation there is but this is weird for whatever country. 

In summation: fuck the justice system. 

[tastefullyoffensive]

Mittens the Texting Cat [via]

[oldbookillustrations]

The moving finger writes; and, having writ, Moves on: nor all thy piety nor wit Shall lure it back to cancel half a line Nor all thy tear wash our a word of it

René Bull, from Rubáiyát of Omar Khayyám, rendered into English by Edward FitzGerald, London, 1913.

A zip file containing the illustrations of the latest series can be downloaded at this link.

(Source: archive.org)

[wilwheaton]

So I’ve been reading about cults for a thing (although cults, as we think of them, and not just harmless hippie communes, are an expression of what’s called a totalist belief system) and I’m reasonably sure the Stupidsphere fits the behaviours, which is worrying because one of the behaviours is ‘stop seeing people who don’t agree with us as people’.   One of the key pieces is having a “milieu” that can be controlled. Fox News and their ilk encourage their watchers not to get information from outsiders, and the American cultural tendency to let people be so long as they’re not hurting anyone keeps outsiders out.   That’s when you start the talk about how it’s imperative that they participate, because they’ll save America from the enemy (defined as: anyone who disagrees with them). They start redefining the world to conform to their ideology (that of outsiders who are simultaneously dangerously effective and incompetent fools). They invite people to share in that ideology, and thereby become heroes fighting the forces of darkness, who are everywhere! Without ever having to leave the couch other than the moments where they blindly vote Republican.   A lot of the techniques that we make fun of Fox News for – their brazen willingness to blend fact and fiction, their use of ‘liberal’ as an epithet, their insistence on impossible standards of ethics – are all techniques to maintain control over their viewership, who are seeing a country their identity is wrapped up in perpetually sliding into the abyss. They’re made vulnerable by their powerlessness, by the Stupidsphere convincing them of their powerlessness.   Fox News is willing to stop here, because all they want is a captive audience, but there’s one further step: when individual experience is superseded by doctrine, you start to encourage erasure of the self, and at that point control is complete. The idea of people, with all their needs and flaws, starts to fall away, replaced by the doctrine, and defending it from outsiders who wish to kill the idea. They are not real; they are flawed, broken, otherwise they would see the perfect truth of the doctrine for themselves. They can not be convinced. They can only be marginalised or destroyed.   Most exit counselling (they don’t do deprogramming any more) starts with teaching the client about methods of control. It’s the same principle as mindfulness – if you can name what’s going on, it’s easier to rise above it.

By reader Merus, on my post “That Time The Stupidsphere Showed Up”. (via wilwheaton)

[bestoffun-blog]

[cardboard-crack]

[markofthemany]

I did not give permission for my likeness and crotchityness to be used in this comic...

[markofthemany]

Just because the monkey can juggle, doesn't mean it understands gravity.

[magicjudge]

So, just to be crystal clear... Haakon, Stromgald Scourge would be the worst EDH general because technically he would start in the command zone from which you could not cast him... yes?

Correct!

[markofthemany]

Booo.

I might just build it anyways for fun casual illegal magic. Thanks for verifying the letter of the law :)

[babyionian-deactivated20200611]

this photoset sums up Magic tournaments so perfectly

[socialmtg-blog]

Well looks like this is spreading here as well, so here is the source and here is a discussion on the Magic subreddit. 

What I want to make sure to say is everyone be aware that you are a representative of the game, and your actions back that up. This wide spread set of images portrays magic players as overweight people who don’t care about hygiene/appearance. It “is EXACTLY how I imagined it." I have friends who play and hide it because they are quickly and harshly judged for it, but this shouldn’t be how it works. Do your best to fight the poor stereotypes the magic community has instead of reinforcing the poor stereotypes. If you aren’t going to fight the stereotype for the sake of your own image, do it for the millions of others who play the game. The guy who did this is clearly just messing around, but it spread like wildfire and is having a poor impact on the millions of players. Think through the consequences of your actions.

[mtgthings]

Please be aware of your actions and how it impacts other people. This does not apply to just magic tournaments, or FNMs, or even kitchen table magic. This applies to life in general. PLEASE.

[markofthemany]

I'm not quite sure if you're speaking to the cracks, the documenters, or those of us reblogging this... I'd like to think it's all of us.

[burritosandpeppermint]

Exclusive Gyms For Members Of Congress Deemed ‘Essential,’ Remain Open During Shutdown

[wilwheaton]

While critical government services are shut down, the exclusive Congressional gyms remain open at taxpayer expense.

Annnnnnd the Democratic-lead Senate scores an own goal. Fucking idiots.

[wasbella102]

[pyrrhiccomedy]

Astronomers have discovered the largest known structure in the universe, a clump of active galactic cores that stretch 4 billion light-years from end to end. The structure is a light quasar group (LQG), a collection of extremely luminous Galactic Nulcei powered by supermassive central black holes.

So that’s cool and everything, but maybe some of you would be interested to know why this is a significant find? Beyond just its record-setting bigness.

Since Einstein, physicists have accepted something called the Cosmological Principle, which states that the universe looks the same everywhere if you view it on a large enough scale. You might find some weird shit over here, and some other freaky shit over there, but if you pull back the camera far enough, you’ll find that same weird and/or freaky shit cropping up over and over again in a fairly regular distribution. This is because the universe is (probably) infinite in size and (we are pretty darn sure) has, and has always had, the same forces acting on it everywhere.

So why is this new LQG so radical? (It stands for ‘Large Quasar Group,’ btw, not ‘Light Quasar Group.’)

Well, let’s try to comprehend the scale we’re dealing with. A ‘megaparsec,’ written Mpc, is about 3.2 million light years long. The Milky Way is about 0.03 Mpc across (or 100,000 light years). The distance between our galaxy and Andromeda, our closest galactic neighbor, is 0.75 Mpc, or 2.5 million light years. LQGs are usually about 200 Mpc across. Assuming a logarithmic distribution of weird shit outliers (if you don’t know how logarithmic distribution curves work, don’t worry about it), cosmologists predicted that nothing in the universe should be more than 370 Mpc across.

This new LQG is 1200 Mpc long. That’s four billion light years. Four BILLION LIGHT YEARS. Just to travel from one side to the other of this one thing. I mean for fuck’s sake, the universe is only about 14 billion years old! How many of these things could there be? 

Right now it looks like the Cosmological Principle might be out the window, unless physicists can find some way to make the existence of this new LQG work with the math (and boy, are they trying). And that’s totally baffling. It would mean—well, we don’t have any idea what it would mean. That the universe isn’t essentially uniform? That some ‘special’ physics apply/applied in some places but not in others? That Something Happened that is totally outside our current ability to understand or quantify stuff happening?

By the way, no one lives there. The radiation from so many quasars would sterilize rock.

Sources: 1 2 3

[markofthemany]

I WAS JUST AT WAL-MART AND THEN

Little boy: Mommy, why is Wonder Woman not wearing any pants?

The Mom: Because Wonder Woman has amazing thighs and she could crush men with them if they insult her.

Little Boy: I wanna be Wonder Woman.

The Mom: Don't we all.

[mtg-realm]

20 sided dice are called D20s, spindowns, or dodecahedra (singular dodecahedron).

These icosahedrons may also be known at ‘Ge3k Bait’ …

[mtgthings]

Spin-down and real D20’s are different.

Spin down (as-per MTG Dice) have the numbers in descending order.

Where D20’s have the numbers evenly spread around the dice.

[markofthemany]

Which theoretically alters their randomness... As you could pick a way to hold and roll the rolldown to achieve a desired result.

Though really I've never even heard of anyone being that hardcore of a magician (cheater).

[finding-things-out-blog]

Schlieren photography

So how does Schlieren photography make pictures like the above?

The phenomena you are seeing is the difference in air density. When gas is heated it expands, and when it cools it contracts. As you might remember from high school physics, when light passes through materials with different densities, it refracts. As light rays pass through a patch of air that is hotter than the surrounding air they get refracted and change their direction. Just like light does when it passes into or out of water. You can see this effect on hot days as the shimmering over roads in the distance.

We tend to think of the air as uniform because we can’t see the tiny changes all around us. But in reality the air is constantly moving and being disturbed by small heat sources. Even your body is causing the air to circulate upward around you.The Schlieren technique greatly magnifies this effect, making even tiny changes visible, such as the heat rising from human skin.

+ Video

Credit: Ian Smith- ian.org