thread for thinkpieces about TEH BIG BAD ALGORITHMS! ALGORITHMS! that don't seem to understand what an algorithm is

quicksort is beautiful: quicksort(x < p) ++ p ++ quicksort(x >= p)

I like thread. And I really enjoy dunking on people writing about the coming gray goo or whatever. Excuse my rambling.

My analysis professor once described learning calculus, like reading Shakespeare or reading a foreign language, as something that makes everyone smarter and see the world in an entirely different way. I always loved that.

I think quicksort could be added to that list of subjects. Everything from the implementation to its analysis and even its peculiar history is complex and fascinating. I actually think a few of these would fit that too. Simplex, Dijkstra’s, or gradient descent would apply too.

If I would’ve posted this five years ago, everyone (myself included) would’ve said gradient descent should probably be removed, right? It’s incredible when ideas can be rediscovered or applied in different ways to transform entire fields. I pay my bills because of gradient descent! If it didn’t come back my research would look completely different.

I really don’t know anything about the two encryption algorithms Diffie-Hellman and RSA outside of the pop science descriptions.

I’m a little surprised by the lack of compiler-related algorithms like LALR or ideas like Chomsky hierarchy of formal grammars or even Church calculi. I first heard about the Kahan algorithm for SVD when I went to a Q&A with Chomsky and someone asked him to share a math trick he liked.

Silby, of the algorithms on the list I’ve implemented, FFT is the trickiest but possibly the most rewarding. It’s usefulness also extends outside of signal processing. I believe NLP, for example, now uses conceptually similar transformations for translation problems.

You should try it! (I bet the key exchange or RSA are trickier but, like I said, I don’t know much about them. Maybe I should try and learn something.)

https://www.fastcodesign.com/90174587/if-you-walk-breathe-or-talk-this-ai-probably-knows-who-you-are

that one's not so bad beyond the headline

this maybe isn't the thread for it but someone I follow on Twitter pointed out Angela Merkel being on the same Coachella tier with an android

What are the implications of AI? How can society interact responsibly with machines? And will our current political strategies help shape the future of intelligent systems? Do not miss out on #moralsandmachines in Berlin on June 27 & 28. Tickets -> https://t.co/s4l6W1aneD #wiwo pic.twitter.com/1GeIBLSKgl

— Léa Steinacker (@leasteinacker) June 7, 2018

Eye in the Sky: Real-time Drone Surveillance System (DSS) for Violent Individuals Identification using ScatterNet Hybrid Deep Learning Network

https://arxiv.org/abs/1806.00746v1

😳

I'm starting to think any paper submitted should require an ethicist be one of the credited contributors before it's allowed to be published

see that doesn't belong in the thread either because it sounds like it actually is a bad algorithm

https://www.theatlantic.com/magazine/archive/2018/10/yuval-noah-harari-technology-tyranny/568330/

I haven't read any of Harari's books, but I feel like the buzz around him is mostly positive so I was surprised to read stuff like this:

Imagine Anna Karenina taking out her smartphone and asking Siri whether she should stay married to Karenin or elope with the dashing Count Vronsky. Or imagine your favorite Shakespeare play with all the crucial decisions made by a Google algorithm. Hamlet and Macbeth would have much more comfortable lives, but what kind of lives would those be?

tbf this is nowhere near as dumb as the OP and he raises plenty of valid concerns, but I'm not sure techno-dystopianism is going to help us anymore than its opposite ever did

The insatiable demand for content and for novelty leads to every kind of inanity getting disseminated. The bar for punditry is set exceptionally low.

harari's 2014 book is compelling, his 2016 one a bit more of a slog

came here to post this

http://www.drb.ie/essays/the-hive-mind

the whole paragraph is a clusterfuck really

Democratic elections and free markets might cease to make sense.

algorithmic trading has existed for over three decades now

(and actually has caused very real and very big problems with the economy, but that isn't as exciting as robots stealing our free will)

only tangentially related to algorithm panic but the best thread on this subject you'll ever erad

(nsfw, probably, conceptually nsfw at least)

Ugh. I will be receiving this article for the foreseeable future, and while I love yelling at people for sending me dumb shit, I have things to do. In an effort to save time, I present

ONE DEGREE OF TRANSLATIONAL FREEDOM DOES NOT A BLOWJOB MAKE

A THREADhttps://t.co/ltHco7EGtU

— Kyle Machulis (@qDot) October 26, 2018

s/read

the links in this thread are like a greatest hits of shitty articles written about THE BIG BAD ALGORITHMS: that awful Racked piece, that terrible "pop music all sounds the same now!!!!!1!1!!!one" study, etc.

https://www.cjr.org/analysis/algorithms-music.php?fbclid=IwAR3gSkSWU0xFnlPK-AkF8BcQwryqYeVdMqB96p_phCE12hyIMuvBbe394Pc

Perhaps with an undertone of personal resentment, phrases like “algorithmic culture” and the “algorithm economy” have cropped up among critics to illustrate the way aesthetic and commercial motivations shift in this world of passive, automated discovery.

TRANSLATION: no one knows what the fuck an algorithm is, and people use "algorithms" as a scary technological bucket to hold their old anxieties about art vs. commerce, authenticity, rockism, and perhaps aging out of technology's target demographic, and I have no idea how people who program things for a living and also listen to music don't want to disintegrate on the regular

“By ‘algorithm culture,’ I meant the notion of art as something reduced to an integer and formula—a constant infinity loop of ‘recommended if you like…’ playlists,” Weiss says.

not to pick on him, I generally like his music writing, but this is kind of hilarious since an infinite loop is, by definition, not an algorithm, which is finite

lol and otm. Weiss might have been gesturing at some kind of cybernetic feedback loop system? Idk I think the discourse/hype around machine learning has made these discussions even worse.

I did think this bit was interesting:

Spotify employs natural-language processing (NLP) models in its recommendation algorithms, analyzing text from blogs, news articles, forums, and other sources to draw connections among different artists and songs, and to figure out what adjectives and moods people associate with these artists online.

Reminds me of people wondering why Conde Nast wanted to buy Reddit. Also that one poster who would freak out about Spotify all the time...maybe he knew

I wonder if they don't actually do that and just say they do b/c it sounds good

a constant infinity loop of ‘recommended if you like…’ playlists,”

If you like Bing Crosby Sings White Christmas you might like Bing Crosby Sings White Christmas... ???

re: the NLP stuff -- I would be skeptical of how useful it is, at least right now. my friend Emily (who actually does know what she is talking about) just posted about some of what's out there, which is... not fantastic: https://emshort.blog/2018/12/11/mailbag-ai-research-on-dialogue-and-story-generation-part-2/

(hey that's a cool person to be friends with)

yeah I wonder if silby isn't right about "NLP". OTOH textual sentiment analysis would be easy enough to run on those sources, though it's pretty dumb imo

oh sorry I see now your friend mentions SA

xp -- mostly through games stuff

I expected exhausting Bandersnatch takes but I did not expect this Jack Thompson twist:

https://qz.com/1513524/black-mirrors-bandersnatch-creates-the-future-not-predicts-it/

But what if instead of logging how many times you watched Love Actually this holiday season, it’s remembering whether you opted to kill your father in cold blood, or save him? What could Netflix do with that highly sensitive emotional information? ... The third concern is the most Black Mirror of them all. It’s not inconceivable to imagine that if the government got a hold of your data, it could think you’re someone worthy of future surveillance. Studies from the Oxford Internet Institute show that there is little evidence to say that playing violent video games lead to violent real-life behavior. However, there are still politicians that peddle this narrative. Could Netflix data be used to identify future terrorists or restrict your access to airports?

excuse me sir, we have here a file from a quote unquote "video game" called The Sims detailing how you trapped your quote unquote "Sims" in a house and set it on fire. sorry, but you have to leave the country now.

something that is actually correct: http://nymag.com/intelligencer/2019/01/facebook-doesnt-need-to-fool-you.html

this in particular should be carved into a goddamn mountain:

In the same way that the breathtaking scope of contemporary surveillance and data-extraction processes makes conspiracy theories about astroturfed memes and bugged smartphones seem almost pathetic in comparison, it also reveals how little our own choices are able to control the flow of our data, and how little our knowledge really matters. I might be aware that photos of myself in 2009 could be misused, and choose not to participate in that meme. But simply by living a fairly regular life on and offline — by clicking on links and writing posts; by opening Instagram and scrolling through it, hovering over some photos and flicking past others; by using credit cards at chain stores; by letting photographs of myself be taken and uploaded to the internet — I’m generating data that’s probably more valuable to the companies involved than those photographs would be. There’s something tragic about the fact that the purely recreational activity of participating in a meme is the subject of conspiratorial paranoia, while the multitude of chore-like activities we do daily, from which data is also being extracted for hoarding or sale, go mostly ignored.

"Can we continue to use your data to tailor ads for you?" Popup irony...

i got stuck at a particular sentence in a blog post recently until i realized that they wrote 'logarithm' in place of 'algorithm'

I expected this bump to be about the aoc thing (which I still need to get through like four levels of Discourse telephone to figure out who's misrepresenting whom and by how much)

i'm not following things too closely but steve bellovin seems cogent

https://arstechnica.com/tech-policy/2019/01/yes-algorithms-can-be-biased-heres-why/

yeah, once I actually read her comments rather than the crust of Discourse coating them, they were far more reasonable than anything in this thread

this one truly has it all https://medium.com/futuresin/why-algorithms-and-artificial-intelligence-are-a-threat-to-democracy-bd4d6b1114af

(I realize picking on something by a "Blockchain Mark Consultant, tech Futurist, prolific writer" is cheating, but it was in my inbox)

IBM must be grateful there's an I in Mafia

ha, I was going to ask, is this actually getting traction? The introduction is disqualifying on its own

the algorithm currently on my shit list is the one medium used to determine, correctly, that I would click this and email me about it

I don't even know

https://www.youtube.com/watch?v=U0F45NHLrRA

https://www.businessinsider.com/hbo-netflix-recommended-by-humans-website-fan-testimonial-2019-8

look out for the algorithims is something

i'm very fascinated by the placement of objects in that room, specifically the large potted plant which almost looks photoshopped in

seriously! like, why is dude putting his tripod on top of the kick drum head? Looks like it could lead to it ripping? ... otoh, he could have just put the tripod on the floor, if it weren't for the immense potted plant that seems to be there for no good reason? The potted plant is really jarring the more I think about it.

also the plant looks a *slight bit* tilted? from left to right? like it seems to tilt more than the bass drum
it's a very claustrophobic mise en scene

yeah ... the plant is really troubling ... maybe it was photoshopped in, but in place of some other object? also the tree branch that seems to be tapping at the window at various times? though we can't hear it in the recording, even though it is ostensibly being recorded through the zoom recorder on the tripod, which would presumably pick up some of that sound? though, maybe not?

and what are the drums resting on? it looks like it's some white pedestal or ... like they are just sitting in mid-air? is it just me, or does it kinda look like the snare/tom stack is just floating there?

really wish I knew who ran this account

Looking for non-algorithmic pop smashes!! https://t.co/SOm0DQqeSQ

— ShittyAandRguy (@shittyAandRguy) December 28, 2019

an otherwise ok aggregation post of a twitter workaround with an unfortunate title: https://www.vice.com/en_us/article/7kz9ez/go-into-2020-by-taking-your-twitter-feed-back-from-the-algorithm

Here's the heart of it:

By adding a few strings—little bits of code used by Twitter to tag types of tweets—to your muted keywords list, you can change the way the Twitter algorithm sends content to your feed.

While logged in to Twitter, go to Settings > Notifications > Muted > Muted words, and add the strings below.

•suggest_activity_tweet: Stops the platform from feeding you tweets you might like

•suggest_recycled_tweet_inline: Stops repeated tweets from appearing over and over

•suggest_pyle_tweet: Stops serving tweets because mutuals engaged with them

•suggest_grouped_tweet_hashtag: Stops tweets associated with popular hashtags from appearing randomly in your timeline

•suggest_who_to_follow: Self-explanatory

•generic-activity-momentsbreaking: Keep tweets served simply because they're part of a Moment out of your feed

right, take your feed back from the algorithm by... adding additional parameters to the algorithm

from this spotty essay on big thief: https://www.dissentmagazine.org/article/listening-in-an-emergency

The task of finding new music presents its own difficulties. If the trouble with YouTube is that it serves up progressively more extreme ideological content, Spotify has precisely the opposite problem, where any song, no matter how brilliant, quickly leads to a chain of forgettable copies that degrade the first song in retrospect.

a) so how, exactly, does this differ from YouTube? because the last time I checked YouTube had a music recommendation algorithm too (although half the time it just sends you to "Plastic Love"). or for that matter, how does this differ from the one on SoundCloud, or Bandcamp, or Amazon, or Last.fm, or literally every other music site that has a similarity feature?

b) 1,000 landfill indie copies exist of virtually every band in existence, because that's how influence works, and the only difference between a playlist and a used-record bin is that the used-record bin probably gave people an advance; in fact one could argue this is what brilliance is, the inability to fade into the background

c) and if Big Thief (the subject of the essay) is indeed brilliant, then why don't they get degraded by the 1,000 Big Thief copies that exist?

d) how exactly does this have anything to do with the rest of the essay

tAngentially

looking at a job advert for someone with a "passion for algorithmic data" to "archive web-based content, including social media data, algorithms, and other important resources", thinking I don't even know what this job advert means or what the job would entail

What even is "algorithmic data"? Does that mean anything, never mind in a sense one could be passionate about? Would the candidate have to email 3l0n to ask for the keys to his Github repository of algorithmic secret sauce so we can print it out and put it in a ring binder?

Anyway as someone who works in something related to archiving, data and algorithms, forced for my degree which I flunked out of to spend inordinate amounts of money on books (either dauntingly thick or insultingly thin for the cover price) with names like "The Derivation of Algorithms", the phrase I think I should know appeared in my head as outraged snark after the question "what even is..." (I don't know anything about anything tbh)

(btw I am actually job hunting now so maybe I SHOULD actually know. Someone tell me lots of things I should know so I can pretend to be employable. Just not in the above-referenced job, for a lot of reasons)

Sounds to me like they are carefully avoiding saying the words "scraping" and "artificial intelligence", as to why they are avoiding these, there could be a number of explanations, I would have a look into the company and see what they say to clients.

You might be right. It's for a university archive where money is scarce (until something a famous author blew their nose on is up for auction) so the answers would probably be "please don't sue us" and "we can't afford to do Big AI Stuff", which is better than my first instinct of "not sure what we're talking about but algorithms are big right now". Or, more charitably, initially written by someone who did know what they were talking about but then went through some managers who wanted to sprinkle buzzwords like glitter and some HR/Comms person who moved words around without knowing the domain.

(And hey, it may actually be worth archiving "algorithms", however you intend to do that, in these weird possibly-last days of 99-100% human devised algorithms before everything gets a little spritz of "we don't know why we pick these search results but we put 'em into our AI black box and it gives these ones more points")

Anyway I'm so old that in my day the big canonical algorithms textbook was called CLR because the letter S hadn't been invented yet CLRS didn't have a 4th author yet, so maybe this thing only sounds weird to me and y'all under-45s are like, get with the programme program app, grandma Karen Becky