recentpopularlog in

robertogreco : compression   16

Teaching is Compression
"Teaching itself comes in at least two forms. It's not just about broadcasting knowledge. No matter how many students you have, if that's all you're doing you aren't making as much progress as you could. The internet is a powerful tool for Type I teaching, but it can't help much with Type II. That is why it is not a satisfactory replacement.

The second type of teaching is a form of compression, making things easier to understand. I don't mean simply eliding details, or making your proofs more terse. I mean compression in the time it takes to explain an idea and its implications.

Computer science is hard. Logic is hard. And that's fine. But if we leave this world as complicated as we found it then we've failed to do our jobs. Think about it this way: if the next generation learns at the same speed as yours, they won't have time to move beyond you. Type II teaching is what enables Type I progress.

Physics went through a period of compression in the middle of the last century. Richard Feynman's reputation wasn't built on discovering new particles or laws of nature, but for discovering better ways to reason about what we already knew. [1] Mathematics has gone though several rebuilding periods. That's why you can pick up a child's math book today and find negative numbers, the square root of two, and many cheerful facts about the square of the hypotenuse. Every one of those mundane ideas was once the hardest problem in the world. My word, people died in arguments over the Pythagorean Theorem. Now we teach it to kids in a half hour. If that's not progress I don't know what is.

So how do we get there in computer science? How can we simplify what we already know so the next crop learns what they need to put our best efforts to shame?

The second form of progress is closely related to the second form of teaching. To my mind, understanding and explaining are just opposite ends of the same process. The only way to prove that you understand something is to explain it to somebody else. Not even using the knowledge is an airtight proof. That's why teachers are always telling you to show your work, to explain step-by-step how you got to the answer.

The most powerful way I know of to understand and explain is through story. Rendering a complex idea into a simple example, analogy, metaphor or allegory simultaneously achieves compression and a way to spread that idea far and wide. Making a good story also forces you to think hard about ways to drive home both the idea and its implications."
teaching  compression  howweteach  explanation  analogy  2012  carlosbueno  richardfeynman  math  mathematics 
september 2016 by robertogreco
WhatsApp Is How Facebook Will Dominate the World | WIRED
"HERE IN NORTH America, mobile Internet traffic is dominated by YouTube and Facebook. So says Sandvine, a company with an unusually good view of the world’s Internet activity. YouTube accounts for nearly 20 percent of all mobile traffic, and Facebook tops 16 percent.

This is what you’d expect. Streaming video from a service like YouTube eats up more network bandwidth than any other type of online application, and in recent years, our smartphones and wireless networks have matured to the point where watching video from a handheld device is a common thing. Facebook is a social networking service, and video is now a primary part of the way people use it.

But the situation elsewhere in the world may surprise you. Take Africa, for instance. In terms of mobile traffic, the continent’s most dominant service is a tool that many in the US haven’t even heard of: WhatsApp.

WhatsApp is the smartphone messaging app Facebook bought for about $22 billion last year, and according to Sandvine—which helps big ISPs monitor and manage all the bits moving across their networks—it accounts for nearly 11 percent of all traffic to and from mobile devices in Africa.

This shows just how popular WhatsApp is across the continent, in large part because it lets people exchange texts without paying big fees to carriers. And it shows that people are using the service for more than just texting. Like other messaging services, it’s a way of trading photos and videos, too. And this year, the company expanded the service so it can make Internet phone calls, echoing services like Skype. According to Dan Deeth—the author of a new report from Sandvine on Internet traffic trends—those high traffic numbers reflect a shift towards voice calling as well as photo and video sharing.

“It’s a mix,” he says. “The texting is the smallest part. Once you get into photos and sending videos to each other and voice calling, that’s when traffic really starts to creep up.”

[image]

Differences in Evolution

In a larger sense, this shows that the Internet is evolving differently in the developing world than it has here in the US. Because network and phone technologies aren’t as mature—and because people have less money to spend on tech—low-bandwidth messaging apps like WhatsApp have become a primary gateway onto the Internet as whole. In Africa, web browsing accounts for 22 percent of mobile traffic, about twice as much as WhatsApp. But no other individual service is even close to WhatsApp’s numbers. Not YouTube. Not BitTorrent. Not Facebook."

[via: "On what makes WhatsApp popular in low-income countries. But the piece overlooks stability. http://www.wired.com/2015/12/new-stats-show-whatsapp-is-how-facebook-will-dominate-the-world/ "
https://twitter.com/anxiaostudio/status/674604771177717761

"WhatsApp is stable and useable under very low/mixed bandwidth conditions. Unlike WeChat and Line it works well on small screens too."
https://twitter.com/anxiaostudio/status/674605226914000896

"Examples re WhatsApp: message queuing when you're offline; low bandwidth mode for voice calls (audio compression)" "@anxiaostudio Wow how do they optimize for the low bandwidth conditions?" https://twitter.com/judemwenda/status/674605980634783745 ""
https://twitter.com/anxiaostudio/status/674608959026675713

"The message queue in WhatsApp shouldn't be overlooked. Most messaging apps give you a permanent error when your note doesn't go through."
https://twitter.com/anxiaostudio/status/674609623236673536

"The little clock next to your note is an assurance from WhatsApp: we'll send this as soon as we can (i.e., you have a connection again)"
https://twitter.com/anxiaostudio/status/674609934135263233 ]
whatsapp  2015  facebook  messaging  mobile  phones  stability  bandwidth  usability  ux  applications  smartphones  connectivity  networking  communication  offline  voicecalls  compression  audiocompression 
december 2015 by robertogreco
Mad Generation Loss - parker higgins dot net
"Mad Generation Loss is a project exploring media encoding and the ways in which imperfect copies can descend into a kind of digital madness. It takes an audio file—here, a recording of Allen Ginsberg reading an excerpt from his seminal poem “Howl”—and adds another layer of mp3 encoding to each second of the sound. That is to say, the first second is encoded directly from the original, the next second is re-encoded from that first lossy copy, and the third encoded again.

[ https://soundcloud.com/thisisparker/mad-generation-loss ]

That sort of re-encoding from lossy originals, known as transcoding, is supposed to be avoided. The generation loss builds on itself, and the quality degrades quickly. That effect is exaggerated here by its second-by-second compounding. By the end of the 3:18 recording, Ginsberg’s voice is nearly impossible to pick out among the background noise.

The last seconds of the recording have been transcoded nearly 200 times. All together, the recording represents nearly 20,000 individual mp3 encodes.

Ginsberg, glitchedThis project takes inspiration from earlier efforts to explore generation loss. “I Am Sitting In A Room” (1969) by Alvin Lucier was perhaps the earliest, and featured a 4-sentence narration recorded on taped, and re-recorded over and over to hear the tape loss. As the narration notes, that process “smooths out” the irregularities of speech, reflecting instead the rhythm and resonant frequencies of the room of the recording.

More recently, an artist named Canzona documented the process of downloading and re-uploading a video to YouTube 1000 times, and the effect of its compound video encoding. He described that project as a tribute to Alvin Lucier.

Unlike those projects, Mad Generation Loss shows the effect of transcoding and loss on a linear recording, not a repeated phrase. The degredation is apparent not from comparing identical inputs and diminished outputs, but from hearing the creep of the telltale white noise and the regular pulse of the mp3s getting stitched together.

The code to create the Mad Generation Loss audio is freely available under the GPLv3. It is written in Ruby and depends on free software like lame, mp3splt, and mp3wrap. Thanks are due to Eric Mill and Ben Gleitzman for technical assistance (though please do not attribute my sloppy code on them), and to Caroline Sinders and Ethan Chiel for their encouragement."
2015  degradation  sound  via:audreywatters  audio  allenginsberg  alvinlucier  canzona  videoencoding  encoding  compression  parkerhiggins  mp3  howl  art 
november 2015 by robertogreco
@reregrammer • Instagram photos and videos
"an iterative Instagram experiment by MN-based artist Patrick Koziol. One image of Alfred Russel Wallace regrammed from the previous."
instagram  degradation  compression  patrickkoziol  photography  digital  instagrams 
october 2015 by robertogreco
The Triumphant Rise of the Shitpic - The Awl
"Let’s call them Shitpics. Because they look like shit.

Shitpics happen when an image is put through some diabolical combination of uploading, screencapping, filtering, cropping, and reuploading. They are particularly popular on Instagram.

For instance, consider this post by the very famous celebrity Ludacris.

[image]

There’s a lot going on here. Let’s try and figure out how this image ended up in its current state.

The image was probably created by the joke account @blackgirlproblems_official, where it looked like this:

[image]

There are a few clues that this is probably the original. The text is centered and sharper, and the emoji is more than a smudge of dirty yellow gibberish. The picture of the monkey is clear (and cute!!!). All of the text in the watermark is legible.

Then this meme went through hell.

It was saved and cropped numerous times. There are a few signifiers of this: The text is cut off on the left side and there are slight black bars at the top and bottom of the frame. The greenish cloud around the text also indicates an absurd amount of (re)compression.

Maybe the most baffling part of this is the appearance of the rule-of-thirds grid, which likely came from Instagram’s upload screen. Which means that someone screencapped their upload process and then uploaded that? And the grid somehow doesn’t even reach the top and bottom edges.

The version of this image from @msrjstlf indicates that it was probably not run through a filter at any point, since the whitespace seems to have stayed mostly that. The lower left corner of the picture does show, however, just how many times it has been reconfigured: the “black” in “blackgirlproblems_official” has been absorbed by section of blanket that has been widening and darkening as the macro travels through the wringer.

[image]

Then Ludacris puts the cherry on top: a translucent gray regram banner crediting the account that he got it from (though not, of course, the original photographer or even macro author).

The Shitpic aesthetic has arisen from two separate though equally influential factors, both of which necessitate screencapping instead of direct downloading. The first is that Instagram, which has no built-in reposting function, doesn’t let users save images directly. This means that the quickest way to save an image on a phone is to screencap it, technically creating a new image.

The second, more important shift is the new macro format that divorces text from image. Classic memes (jfc “classic memes” what are we doing) had text directly on the image, written in Impact font in a particular style—white with a black border. That changed with the rise of the text setup/image punchline format on Tumblr, particularly on the blog What Should We Call Me, which spawned and continues to spawn imitators. Twitter began to imitate this when it changed tweet formatting to hide image URLs (pic.twitter.com) from tweets, easing the transition from text to image, from setup to punchline.

It’s difficult to send someone a technically exact copy of these types of jokes, because they can’t be bundled into a single file such an image. Sending the URL where the joke is hosted requires someone to load an entire webpage, which is relatively laborious on mobile, and so they necessitate being screencapped.

In general, directly saving images on mobile is a function that, even when available, most people don’t bother to use or even learn (saving files locally—in any kind of file system—is generally discouraged in smartphone operating systems). Screencapping is just easier—it’s the quickest way to get something from the internet to your camera roll. That’s why even classic-format memes have fallen victim to the Shitpic process.

[images]

When you pair the format’s inherent need to be screencapped in order to attain virality with Instagram’s prevention of downloading images, you get an endless cycle of screencapping and compression through uploading. Throw in the occasional filter, or watermark, or regram tag, and let the process carry itself out for a while, and eventually you get a Shitpic. The layers pile up, burying and distorting the original.

The rise of the Shitpic demonstrates just how little ownership there is on the internet: Shoddy workarounds and subpar image quality are a small sacrifice to make, so long as your version of a joke goes viral instead of someone else’s. That the image is a muddled cacophony of compression artifacts and blurry emoji matters little, so long as your screenname is above it.

[image]

Perhaps most importantly, the Shitpic aesthetic could very well be the first non-numeric indicator of viral dissemination. Metrics such as pageviews, impressions, Facebook referrals, YouTube view counts, and BuzzFeed viral lift all attempt to quantify virality in some way. To the layman (and, let’s be frank, some industry experts too) all of this is gibberish.

[images]

But if you look at a Shitpic, you can instantly tell the level of virality by how worn it looks, how legible its text is, how many watermarks adorn it. You can count them much like you would rings on a tree. A pristine-looking meme engenders skepticism—“This can’t be that funny, it hasn’t been imperfectly replicated enough.” But when you see that blurry text, partially cut off by the top of the frame, and a heavily compressed picture of Kermit below… that’s when you know:

This is gonna be a good-ass meme."
instagram  photography  internet  culture  degradation  compression  cropping  2014  brianfeldman  digital  shitpics  mobile  phones  screencapping  screenshots  distortion  virality 
october 2015 by robertogreco
Generation loss - Wikipedia
"Generation loss refers to the loss of quality between subsequent copies or transcodes of data. Anything that reduces the quality of the representation when copying, and would cause further reduction in quality on making a copy of the copy, can be considered a form of generation loss. File size increases are a common result of generation loss, as the introduction of artifacts may actually increase the entropy of the data through each generation."
degradation  audio  video  photography  photocopies  analog  digital  compression 
october 2015 by robertogreco
Katherine Hayles - A Theory of the Total Archive - YouTube
"Katherine Hayles - A Theory of the Total Archive: Infinite Expansion, Infinite Compression, and Apparatuses of Control"
books  archives  borges  christianbök  shipoftheseus  longevity  lifelogging  2015  badrobotproductions  jjabrams  katherinehayles  libraries  communication  recovery  memory  expansion  compression  control  words  srg  s.  dougdorst 
april 2015 by robertogreco
The Conscientiousness of Kidspeak - The New Yorker
"Often enough, something we propose as a serious idea turns out to be more or less a joke. It’s much rarer that something proposed as a joke—or, at least, proposed as a semi-serious conceit, offered in the spirit of what’s often called, grimly, “tongue in cheek”—turns out to be, or to have the germ of, a serious idea. So I was startled and delighted the other morning to find out that a small joke I made a few years ago turns out to be true (or true-ish, anyway) and can be shown to be so by a recent scientific (or scientific-ish) paper. It started when, in 2011, I was writing about attempts to computerize the translation of natural language. I touched on the omnipresence of “like” and similar verbal tics in Kidspeak—the language of twelve- to fourteen-year-olds, particularly girls—a dialect about which I have what social scientists refer to as “a strong informant” right here at home. The ubiquitous qualifiers in this dialect—the constant “um”s, the continual “you know”s, and, above all, the unending stream of “like”s—are, it’s usually said, a barrier in the way of lucidity, brevity, and making a point.

But, as I wrote then, we’re all naturally quite good at compressed, or telegraphic, speech, where what is omitted is implicitly understood by the listener. For the sake of economy, we have to leave a lot of information out of everything we say, and one of our special human abilities is to make that economy itself eloquent and informative. Kidspeak is a classic instance of compression in balance with concision. What sounds limited and repetitive to the outsider is, to the knowing listener, as nuanced as a Henry James passage.

If, for instance, a fourteen-year-old girl says, “So we, like, um, went to the pizza place, but the, uh, you know—the guy?—said, like, no, so we were, like, O.K., so we, uh, decided that we’d go to, like, a coffee shop, but, uh, Colette can’t—she has, like, a gluten thing. You know what I mean? So that’s, like, why we came home, and, um, you know, would you, like, make us eggs?” To a sensitized listener, who recognizes the meaning of the circumlocutions, the nuanced space between language and event, the sentence really means: “So we tried, as it were, to go and enjoy a pizza, but the, so to speak, maître d’ of the establishment claimed—a statement that we were in no social position to dispute—that there was, so to speak, ‘no room for us at the inn.’ And then Colette insisted—and far be it for me either to contest or endorse her self-diagnosis—that she could not eat wheat-based food, so, knowing full well that it is likely to be irksome and ill-timed, could you feed us with scrambled eggs?” The point of the “likes”s and other tics is to supply the information that there is a lot more information not being offered, and that the whole thing is held at a certain circumspect remove. It didn’t happen exactly this way, and, of course, one might quibble with a detail here or there, but this is the gist of what happened. Each “like” is a Jamesian “as it were.”

It turns out that three sociolinguists at the University of Texas at Austin have been studying these things systematically. The paper they produced, published in the Journal of Language and Social Psychology, has the beautiful title “Um … Who Like Says You Know: Filler Word Use as a Function of Age, Gender and Personality.” The study they conducted “aimed to investigate how the frequency of filled pauses and discourse markers used in the English language varies with two basic demographic variables (gender and age) and personality traits.” The researchers explain that, to do this, they “focused on three common discourse markers … (I mean, you know, and like) and two filled pauses (uh and um).”

They recorded and transcribed interviews with the speakers, noted how often the speakers used so-called “discourse markers,” and concluded that these markers are, indeed, used most frequently by women and girls. More important, the study also shows that the use of the discourse markers is particularly common among speakers who score on a personality test as “conscientious”—“people who are more thoughtful and aware of themselves and their surroundings.” Discourse markers, far from being opaque, automatic, or zombie-like, show that the speaker has “a desire to share or rephrase opinions to recipients.” In other words, those “like”s are being used to register that what’s being narrated may not be utterly faithful to each detail—that it may not be, as a fourteen-year-old might say, “literally” true—but that it is essentially true, and, what’s more, that an innate sense of conscientiousness and empathy with the listener forbids the speaker from pretending to a more closely tuned accuracy than she in fact possesses. As one commenter on the paper writes,
The researchers believe the explanation is that “conscientious people are generally more thoughtful and aware of themselves and their surroundings,” and their use of discourse markers shows they have a “desire to share or rephrase opinions to recipients.” Stated slightly differently, discourse fillers are a sign of more considered speech, and so it makes sense that conscientious people use them more often.

So it seems that the conscientiousness of “like” is what makes it appear so often. All of the circumlocutions of Kidspeak underline not sloppy indifference but undue scrupulousness. We should admire, not belittle, kids who use it. Far from being banished from polite or public dialogue, their discourse markers should mark our own—they should be imported as a sign of a meticulous grasp of the truth that there is no settled truth, that all narration is subjective, that every account must always be qualified. A headline in the Times, to be so, might read: “SCALIA, LIKE, SAYS THAT OBAMA, IS, YOU KNOW? LIKE, NOT COOL, BUT, O.K., DO IT. WHATEVER.” If the people at the Times wanted to run a truly conscientious newspaper, anyway, they would.""
language  conscientiousness  adamgopnik  2014  kidspeak  awareness  discourse  empathy  thoughtfulness  fillerwords  communication  like  research  linguistics  brevity  lucidity  compression  concision  henryjames 
july 2014 by robertogreco
Panic Blog » ShrinkIt 1.1
"ShrinkIt is a simple, small, Panic-internal tool (for Mac OS X Snow Leopard) that will automate the process of stripping needless metadata from PDFs by re-saving them using Apple’s PDF processor. For app resources and icons that aren’t using high-end Illustrator features, this should be lossless — Apple’s PDF code is not compressing anything, just removing cruft. Simply drop a bunch of files (not folders) onto it — such as the contents of your app’s Resources folder — to have it find the PDFs and do its magic. The original files will be renamed with the prefix “_org_” for backup safety. That’s it!"
shrinkit  adobe  mac  osx  optimization  utilities  pdf  freeware  panic  software  applications  macosx  compression  free 
february 2010 by robertogreco
Like a Wal-Mart shirt with hand-sewn seams « Snarkmarket
"Usu­ally, we expect quan­tity to com­pete with qual­ity. You know, like: YouTube’s all about vol­ume; Vimeo’s all about qual­ity. That’s the break­down that we expect. Cheap and mass-produced vs. high-end and artisanal. Except that YouTube is the qual­ity leader, too...just talk­ing about tech­ni­cal qual­ity, of course, and Vimeo has done a great job cul­ti­vat­ing qual­ity of a dif­fer­ent sort...here’s a tip for video uploads from the iPhone 3GS: Don’t send the video directly from the Cam­era app. It com­presses it severely, and there’s no way to tell it not to. Instead, copy the video from the Cam­era app, then open Mail, cre­ate a new mes­sage (to your YouTube upload-via-email address) and paste the video in. Voila. No com­pres­sion. Of course, the video takes com­men­su­rately longer to upload, but it’s well worth it. The same trick works for photos."
iphone  video  youtube  vimeo  compression  quality  quantity 
november 2009 by robertogreco
smush it!
"Image optimization is an art that not many people master. There are many good image editing tools that allow us to get the best visual result for a certain file size but "under the hood" a lot more optimization can be done. Smushit.com is a service that goes beyond the limitations of Photoshop, Fireworks & Co. It uses image format specific non-lossy image optimization tools to squeeze the last bytes out of your images - without changing their look or visual quality. You'll get a report of how many bytes you can save by optimizing your images and all the changed images as a single zip for download."
webtools  onlinetoolkit  images  imageoptimization  yahoo  compression  optimization  webdesign  webdev  technology  performance 
october 2008 by robertogreco
buffington. Songs and how miserable my brain is with them...
"I've been thinking a lot about the brain and how memories are retained; this was a perfect example of how "compression" works in the brain. Rather than store raw, high definition audio, it stored concepts about a single song."
memory  brain  music  learning  compression 
april 2008 by robertogreco

Copy this bookmark:





to read