Replying to @fatemehx2
I would recommend people log what proportion of their users have DNT on. They'll be surprised, even with Chrome where it's hidden deep.
Oh, and some trackers abide. IIRC BlueKai does die instance.
March 2019
150 tweets
Is there a (respectful) way of downloading the whole SEP? (I'm already a friend.)
Still waiting for `line-break: yolo;`
Quoting a tweet by @intenttoship ↗
Replying to @taravancil and @anildash
Indeed, our regular wintry mix is more about the snow but we figured that throwing some rain in between two squalls would feel, well, warmer.
♥ 1
Replying to @skdh and @seanmcarroll
Sure, but reverse search also means that advertisers who have your number (maybe because you had to provide it for a delivery) can now use it to target you. Using a number given for security purposes for anything else, let alone ads, is a betrayal of trust.
Furthermore, this hurts trust that people have in 2FA setups. It creates social harm.
Replying to @simevidas
line-break: HAMMERTIME;
What cracks me up is that if you showed ads at random you would probably reach women 24-55 more than the 20-30% success rate than they saw.
If you used contextual targeting you would definitely hit much higher for much less money…
Quoting a tweet by @Digiday ↗
♥ 1↺ 1
I would absolutely watch a show that’s nothing but @AOC asking the most relevant questions, all day, every day.
Quoting a tweet by @Public_Citizen ↗
♥ 28↺ 7
The worst part is that there's a bit in my brain that can't help wanting to read the part about Monsters Inc. as a child-eating tutorial.
Quoting a tweet by @oneunderscore__ ↗
♥ 4↺ 1
Replying to @jason_kint
Did too! But because I want to know how to eat children (mine would fight back too hard) but because I'm curious about how you even make that argument sound like something even half cogent using, like, words and perhaps even sentences.
♥ 1
"Time and Chance", by David Albert. An enchanting exploration of the foundations of statistical mechanics. It took me a little while to get used to his style (and his incorrigibly Cartesian love of going down blind alleys to make a point) but I grew to really enjoy it.
♥ 2
Replying to @sandhawke
This one is a short book, 170p. It’s a slow read, though, since it requires concentration. I do that in commutes and evenings (though I fallow books around). I guess the only trick is public transportation and no TV?
♥ 21
Replying to @sandhawke
Oh, yeah: you also need to put your device away. And account for the fact that it’s like exercise: it’s a practice to build up towards!
♥ 1
Replying to @null and @dauwhe
It is! It’s sort of the gateway drug to emergence in physics.
Replying to @null and @dauwhe
Hahaha! It’s too powerful even for grown scientists :)
Replying to @_abialas and @femalegazebot, @can
I love them! It’s like some demo scene kids retired and now have to turn tricks to pay their loan shark!
♥ 1
People keep saying that UX is make-or-break but how then do you explain the survival of Venmo, Jira, Apple Finder, Facebook, Amazon, Netflix, or American democracy?
♥ 18↺ 3
Replying to @null and @amyvdh
Legit.
Replying to @stimpee
We use SNow and indeed it's clunky and slow, but then so is Jira. At least in SNow the text entry widget isn't useless.
Replying to @ubiquitic
I've clicked on a lot of things there, usually because it's not obvious how to do something very advanced like closing an issue. I've never had a usable experience as a result...
Replying to @_abialas and @femalegazebot
I was trying to find the fave button for this tweet when...
♥ 2
This is not @SenWarren just railing against Big Tech to score points: it’s a much, much needed policy proposal to bring anti-trust back to life after decades of the US giving up on having a free market economy. nytimes.com/2019/03/08/us/… (via @MaxGendler).
♥ 6↺ 3
What if Child Care Were as Standard as Coffee at Tech Conferences?
@natalyathree is absolutely right here in her @timesopen article.
open.nytimes.com/what-if-child-…
♥ 3↺ 2
Replying to @JonFerraiolo and @SenWarren, @MaxGendler
I highly recommend reading @superwuster’s “The Curse of Bigness” on this topic. Consumer harm can be very hard to prove and it’s not the only kind of harm that a large company can make. There is also substantial evidence of good coming from break ups.
Replying to @JonFerraiolo and @SenWarren, @MaxGendler, @superwuster
I’m not sure we’d have had a Microsoft process if the IBM one hadn’t happened; we almost certainly would not have the issues we have today with Google and Facebook if Microsoft hadn’t been sued out of its position because they wouldn’t exist.
Replying to @JonFerraiolo and @SenWarren, @MaxGendler, @superwuster
That if Microsoft had no been sued (in the US and EU) they would have prevented the emergence of Google and Facebook. If IBM had not been sued they would have prevented the emergence of Microsoft.
It’s just good forestry.
Replying to @DerekSkillings
Ah, wish I'd known you were speaking! I'm not very far, would've loved to come listen.
♥ 1
Replying to @fatemehx2 and @eiannopollo
Overall I agree with the threat (though given this is Facebook I would expect more from government than from governance), but I don't understand why that would remove rights to be forgotten? It's not as if Facebook has used any meaningful type of consent...
Replying to @thom4parisot and @oncletom, @Thierry, @nhoizey
Pas mieux.
Replying to @null and @SimonDeDeo
If won't be poisonous, and the odds are it's good. You'll know fast!
This house has all the best books.
♥ 4
The primary purpose of commit messages is to give finicky colleagues something to do to prevent them from interfering with anything of actual substance.
Quoting a tweet by @Amycruz0x ↗
♥ 1
Replying to @IrisClasson
I used to keep a folder with "interesting code" printed out.
Replying to @frivoal
Where do you sync to?
Replying to @torgo
I hope you are pushing for "hyeef"?
Replying to @frivoal
Dropbox is not the culprit. But sometimes pics get synced in several ways. Media and storage permissions are pretty broad, it could be a source. At this stage it's still a coincidence; maybe make some tests with other objects!
Replying to @sandhawke
Crafting detailed commits and reading those of others.
Replying to @LogicalAnalysis and @Noahpinion
I don't know if it's impatience, I wonder if it isn't boredom possibly in part due to the fact that very little of this is curiosity-driven tinkering. Much of this you can't do in your garage, you need state-like access to data and teams that's the province of megacorps.
♥ 1
I hope other parents in DST areas are enjoying the annual Dawn of the Grumpy Zombie Children as much as I am.
♥ 4
Replying to @sandhawke
Some people can have endless debates about these so it would seem life yes!
Replying to @frivoal
That's an improvement!
Replying to @dontclickonthat
Oh yeah. My 5yo's teacher is pregnant, too...
When Tech does Ag, Food, or other real things in the same way they do, well, tech what could possibly go wrong?
Quoting a tweet by @SarahTaber_bww ↗
Replying to @theturinghorse and @sindarknave, @SimonDeDeo, @KentKChang, @Ted_Underwood
Yeah, what Turing said.
♥ 1
The @nytimes Data team just DROPped an article explaining how we hire: “A Window Into Our SQL Interviews: How We SELECT Data Analysts At The New York Times” (feat. @MaxGendler).
And yes — we hire for sense of humour.
open.nytimes.com/a-window-into-…
♥ 4↺ 1
"Counterfactuals and Causal Inference", by Stephen Morgan & Christopher Winship. Delves into quite some depth of detail on the topic but always in a clear and didactic manner.
♥ 3
Amongst the many perks of working with wonderful people: getting a unicorn cake!
🦄 🎂 ♥
♥ 17
Replying to @fatemehx2
Just imagine what the next three months could bring!
Replying to @null and @amyvdh
Yes it is! Thank you 😘 When are you coming to NYC? It's been forever...
Replying to @Cennydd
It's also going about it in a very idiosyncratic way. I remember that web, it wasn't like that.
Same. All I ever watch on @YouTube are trailers and the occasional science or philosophy talk, yet my recommendations often tell me that Obama is a devil-worshipping Illuminati.
It's an interesting question how much of Google's revenue is profiting from harm to democracy.
Quoting a tweet by @maplecocaine ↗
♥ 3↺ 1
Replying to @OriPekelman
Merci 🤗
Replying to @slightlylate and @IKEA
Something fiery, renegade, with disruptive, even disturbing copy.
♥ 1
Does anyone have or maintain a convenient list of security compromises that relied on the fact that sites routinely embed third-party code injection for tracking and advertising purposes? And also for SDK embedding in mobile apps?
♥ 9↺ 9
Replying to @cyberbaloo_
No, but we do have a great baking club
Replying to @TheAnnaGat and @can
It can evolve your intellectual life in the same way that mass extinctions evolve life on Earth 🌍.
Replying to @torgo and @inboxbygmail
Is the internet broken? Someone should have suggested "pine" by now.
*taps glass*
IS THIS THING ON?
♥ 1
Replying to @kottke and @monteiro
This looks like it's Kindle-only, any chance of it becoming available on something that designers haven't destroyed, like, say, paper?
Replying to @skdh
I'm sorry but this is bad reporting. These platforms have tools to remove content at scale: it's what they do for copyright infringement. It doesn't matter if it's two uploads or millions, perceptual matching works just the same.
The bar is: did they act as fast as they would have for, say, the latest GoT episode leaking online. If yes, fine, if not...
♥ 1
Replying to @monteiro and @kottke
Wonderful thanks!
Replying to @versac
De dons, certes, mais aussi de données. Je pense que les trackers sont mis en place avant la création du site 😋
There is nothing controversial about that. The primary reason that targeted advertising persists is because marketers won't get fired for using it, not because it's cost-effective.
Quoting a tweet by @mmasnick ↗
♥ 6↺ 2
Replying to @csarven and @danbri
I don't know man, I'm not in this space at all anymore, and I'm not in touch with the science.ai folks.
♥ 1
Quantum physics takes you new places.
♥ 1↺ 1
Les lois ne font plus les hommes mais quelques hommes font la loi. #ChansonVérité
♥ 1
Replying to @LAM_Barrett
♥ 1
Replying to @tobie and @w3c, @jeff_jaffe
I want to buy the same clothes you're wearing now!
Replying to @tobie and @w3c, @jeff_jaffe
Because you're an influencer. At least, I've heard that's how it works.
♥ 1
Replying to @tobie and @w3c, @jeff_jaffe
Currently dressed for sports, not sure that's a win for you 😁
Replying to @Chronotope
It's not too late to push back. Part of the problem is that publishers have not picked up a place to set standards. We could work on that as a group. Cc @jason_kint @wseltzer.
♥ 2↺ 1
Browser friends: how much of the web breaks (other than ads) if document.write in an iframe becomes a noop? Same with eval (we could make a list).
Quoting a tweet by @Chronotope ↗
♥ 12↺ 3
Replying to @jaffathecake
Niiiice. I believe Yoav had mentioned this but I forgot, thanks! Can we have that on script?
Replying to @triblondon and @jaffathecake
We should definitely talk! I'm soon to look into seeing if we can do some brutal things to misbehaving ads with CSP so this would fit right in. Do you have my work email?
Replying to @jaffathecake
Almost all third-party stuff is done through code injection. There are far bigger issues than defer can hope to solve and it would be great to be able to limit what they can do. I've been wondering about a SafeWorker or some such.
Replying to @jaffathecake
I'm totally out of the web loop so I might be missing something new, but I don't think so here.
Replying to @larsberg_ and @ServoDev, @lastontheboat, @nokusu
There is no good reason for ads to use that! It's fine if they fail 😊
Replying to @nokusu and @larsberg_, @ServoDev, @lastontheboat
Originally I was looking for it as a noop, not an exception.
Replying to @hsivonen
Do they have (today) a defensible reason to load script this way?
Replying to @jaffathecake
It’s more that it’s exposing too much, notably fingerprinting APIs. I’m looking for a context that can do little more than read a bit of data and send that home.
The use case is: today publishers need to integrate with third parties, but they’re only so trustworthy…
♥ 1
Replying to @jaffathecake
Ideally I’d want to load them into an environment I control so that they can only do what they’re supposed to do. No fingerprinting, perhaps as far as a fake `document` to make them thing everything’s normal and feed them controlled information.
Replying to @jaffathecake
I could have my own Worker set things up that way, but `importScripts` would need them to cooperate with CORS so I don’t think it works.
Replying to @hsivonen
I know, but ads don't typically load dependencies, they load trackers that are usually independent of one another.
Replying to @LogicalAnalysis
What I find missing in those discussions are impediments to bringing it back. Is it only a collective action issue? A small coordinated group cuttle break out of that.
♥ 1
Replying to @tabatkins and @hsivonen
Yes, but this goes back to my initial question as to whether there are many cases of d.w in iframes. If you think there are, then indeed it's worth breaking down the different motivations.
Ads and trackers. It would be interesting to make them declarative so that users and publishers could have enforceable policies beyond yes or no. A good thing to hack on with custom elements! t.co/uLf2Id6oju
♥ 5↺ 2
Replying to @hsivonen and @tabatkins
All I'm saying is that from the sample of the world I see that's at best a minority case and shouldn't be needed with modules. I'm totally open to being wrong.
By volume, scripts in iframes are certainly used far more for bad than for good though 😑
Economist friends: has anyone studied the effects of a monopsony on a market of non-excludable goods (like data)? (It is maintained through a monopoly on the other side of a two-sided market.)
♥ 1↺ 1
Over the past weeks I’ve divided my time between working to fix digital privacy, reading about the philosophy of quantum foundations, and trying to get @IKEAUSA to deliver a couch.
It’s not immediately obvious which of the three presents the hardest problem.
♥ 12↺ 1
Replying to @trekonomics
Do you write in GDocs? I’m curious about good options in this space, especially for non-trivial, long documents.
♥ 1
Replying to @trekonomics
I’ve found it to become slow over ~40 pages, but that may also be because of the number fo comments.
♥ 1
Replying to @shimown and @IKEAUSA
Right. In the Stockholm interpretation a particle can register as delivered without having actually reached the detector.
♥ 3
Replying to @DailyNousEditor and @seanmcarroll
I hate to bug you but it's "Châtelet", with a circumflex accent.
♥ 1
Replying to @andreisambra and @Fnac
Hahaha, ça fait bientôt 10 ans que j'essaie de quitter leur mailing Oxybul.
"Beyond the Dynamical Universe", by Silverstein, Stuckey, and McDevitt. This seemed... handwavy? I stopped halfway through because they mention lots of interesting problems but it felt like they didn't ever get to very crisp answers. I'd be curious to hear other opinions.
Replying to @cramforce
Slow ads can still lose money!
Replying to @cramforce
Is there really a difference between 2G and a modern news site? 🙄
Replying to @slekies
Thanks, very helpful, I'll dig more there.
Replying to @IKEAUSAHelp
You know, for that to work there has to actually be someone answering the DM, right?
Here's how to emulate @IKEAUSA support:
1. Start with a can-do attitude;
2. Apologise like you're Mark Zuckerberg;
3. Promise to get right on it;
4. tumbleweed.gif...
Quoting a tweet by @IKEAUSAHelp ↗
♥ 3
Replying to @annaeveryday and @Floridi
Not just @Floridi, but also @j2bryson and @ssnstudy. I somehow don't think they would expect to be able to provide valuable input in this context?
♥ 4
Replying to @j2bryson and @annaeveryday, @Floridi, @ssnstudy
I would argue that that is heavily context-dependent. If running a country you might have to talk to homicidal maniacs. In tech ethics you might need to engage with the IAB. I'm less certain that responsible AI benefits from engaging with a transphobic racist?
♥ 4↺ 1
Replying to @j2bryson and @Floridi, @ssnstudy
I am not immediately convinced that global governance involves talking to people who think global warming is a hoax.
Replying to @lastpositivist and @yannick___m, @briandavidearp
No, incognito should still work... 🙄
“Beware the DAG!”, by A. Philip Dawid. A critical look at assumed relationships between conditional independence and causal models, and a description of an approach with a more explicit delineation.
proceedings.mlr.press/v6/dawid10a/da…
♥ 2
I thought that this was a clear and illuminating paper, I am just slightly disappointed that it wasn’t called “Man Bites DAG!”.
♥ 3
Replying to @null and @SimonDeDeo
Napoleon did in fact use A4 as far as I know! It's one part of the Revolutionary metric stuff.
Replying to @reidmore_online and @wendy_a_reid
It's 100% normal. It's also not just the first time...
♥ 1
This is such a simple and such a good idea for so many reasons that I'm going to need several tweets to tell you how good it is and why we should do it right away. 👇
Quoting a tweet by @pnhoward ↗
♥ 20↺ 10
If you're too lazy to read the article, the idea is to get all ads, worldwide, all the time, into a common public archive. At times the article seems to focus on social media ads but it should really be all.
♥ 2
In case you're wondering: yes, that's a lot. But a tiny fraction of what is lost to ad fraud could pay for @brewster_kahle and the @internetarchive to run it. Publishers would not run ads unless they had an integrity hash matching that public record. It would be illegal.
♥ 3↺ 1
Replying to @robleathern
I'm glad to hear that you're working on improving it! I know previous versions had performance issues, is it better now?
I know you auto-classify but as you know that has its own problems. An archive of everything that others can classify in their way is more transparent.
Replying to @robleathern
I'm also curious to hear your thoughts about how to expose to users what targeting was used. Last I looked, Facebook users could see who had targeted them with an audience created from hashed emails but couldn't see Pixel-based audiences.
This could be seen as shaming brands away from using the much more privacy-safe uploads and towards more problematic tracking. It would be great to see those placed on par.
Replying to @robleathern
I will, thanks!
Replying to @ProgrammerDude
We could pretty much use SRI!
Replying to @dmarti and @robleathern
I don't claim that it is 100% privacy-safe, only that it is much safer than Web trackers. Also, in Facebook’s case, they are controllers of tracker data but only processors of uploads. That's a huge difference to user privacy.
Replying to @dmarti and @robleathern
Now of course you can cheat, but that seems more like spam than privacy? Also, cookie syncing often does similar things.
Replying to @dmarti and @robleathern
I really don't want anyone to do the latter. It doesn't work. Legitimate news gets flagged as political. This doubly so in a crazy world in which antivax or climate denial are somehow considered legitimate political positions.
Replying to @dmarti and @robleathern
Calling bullshit shouldn't be a political stance, but under such a regime it is.
They could do much more to police the first though. It would be interesting to see how a culture shift towards governance and a renewed focus on products over measurement could change Facebook.
Replying to @robleathern and @dmarti
I know, but that just pushes the issue off to figuring out what is news. That's why I like the idea of just publicly archiving all ads and letting researchers and journalists find the problems.
Replying to @sandhawke
Rabbits, for instance, are somewhat a duck.
Replying to @dmarti and @robleathern, @pnhoward
That's the hard part. You can easily include the platform's targeting parameters if they exist but how can you safely capture segments built off the platform that are then just identifiers without compromising the privacy of those people?
Replying to @dmarti and @robleathern, @pnhoward
There is much more in the world than custom audiences, though. And given the scale (and the risk) I don't think consent can work.
Replying to @dmarti and @robleathern, @pnhoward
Maybe there's some kind of smart trick with Bloom filters that could work.
"Philosophy of Physics: Space and Time", by Tim Maudlin. An approach grounded in geometry, progressing effortlessly from Aristotelian space to General Relativity with impressive clarity. I look forward to reading volume 2.
♥ 1
Replying to @null and @SimonDeDeo
Those two options aren't mutually exclusive.