Andrew Yang has released a plan for Internet regulation. Some of it is good, some of it is anodyne, and some of it is terrible. You may wonder why I’ve chosen to write about this, since he’s not going to be the nominee. Two reasons: one, we should all be talking about this stuff more. Two, his plan is a great example of how Americans go astray when thinking about regulating tech companies.
From his website, we can start with the good:
Regulate the use of data and privacy by establishing data as a property right. The associated rights will enable individuals to retain ownership and share in the economic value generated by their data.
Wonderful! Big fan. I used to work at a startup that sought to enable customers to do this. Alas, it failed because investors thought that our user base would only be the Very Online™, and the sorts of people who want to hide things. Without widespread adoption (or government regulation), this doesn’t work.
Then we have the “sure, alright, I could see that working out”:
Minimize health impacts of modern tech on our people, particularly our children. I will create a Department of the Attention Economy that focuses on smartphones, social media, gaming, and apps, and how to responsibly design and use them, including age restrictions and guidelines.
He proposes things like “removing autoplay video for children under 16,” though personally I’d like to see that apply to children under 116. This one also contains stupider bullet points like “removing the queues that allow infinite scrolling,” which would be laughably easy to work around, if anybody can even figure out what it means.
And then we reach his dangerous ideas, which many politicians across the spectrum like, and which represent a particular failure mode in American regulatory thought.
There are essentially two ideas united by the same fallacy:
Social media platforms have catalyzed mass disinformation campaigns over the past decade, threatening not just our wellbeing but our democracy… Algorithms driving recommendations towards conspiracy theory content or other types of disinformation need to be reined in…
We must address once and for all the publisher vs. platform grey area that tech companies have lived in for years. Facebook, Twitter, and other social media sites are using algorithms to make recommendations. These recommendations drive the majority of traffic, up to 70% for Google owned YouTube.
Section 230 of the Communications Decency Act absolves platforms from all responsibility for any content published on them. However, given the role of recommendation algorithms—which push negative, polarizing, and false content to maximize engagement—there needs to be some accountability.
I will leave aside the fact that Section 230 does no such thing and write instead about this inane focus on recommendation algorithms. (If you’re curious about my opinions on Section 230, there are upcoming EU regulations that touch on this, which I have written about here.)
Yang notes that large social networking sites are used to push massive amounts of misinformation. Further, such sites use recommendation algorithms that often reward bad-faith actors. Since these sites are so large and ubiquitous, this represents a threat to consumers and democracy. So far this is not controversial; even Facebook and Twitter acknowledge this.
However, Yang’s proposed remedy is to require said recommendation algorithms to be either posted in full online or pre-approved by the government. The former would needlessly stifle expression and innovation, and as for the latter, well, imagine Trump with that power.
So where is this faulty reasoning I’m talking about? It’s very simple. If there’s a tech company that’s so big and ubiquitous you feel it’s a danger to democracy, what you have isn’t a justification for an algorithm-policing bureaucracy—what you have is a company that’s too big. And the government already has the authority to address that.
Yang says some dumb shit about how antitrust won’t work in his proposal, which demonstrates a follow-up point: The concept of antitrust enforcement is so foreign to modern Americans that they often reach for bizarre & harmful over-regulation to mitigate the damages caused by huge companies instead of just banning huge companies.
While Yang is a sideshow, and I suspect he does not understand what a recommendation algorithm is, this mode of thinking is very common, so I wanted to start a conversation. I’m happy to get into more detail in the comments.
As a 25-year veteran tech professional with a patent and quite a few products to his name, I don’t think posting algorithms online would be needlessly stifling to innovation any more than I think that posting patent details is. However, I agree with the rest of what you’re saying.
uh, okay then.
Why do people take anything Andrew Yang says about technology seriously? Read his Wikipedia page – there is little to nothing in his background that suggests he knows much, if anything, about technology. I honestly think people just assume he knows what he is talking about re: technology because of ethnic stereotyping (which he himself abets by selling ‘MATH!’ hats).
Major Major Major Major
@Adam Lang: Fair. I read “open source” and interpreted it as “ceding rights,” but you’re correct that it’s not necessarily the case—although proving infringement when the thief doesn’t share their own code is impossible.
This is a actually not a great model. You can alienate your property freely and permanently. You don’t want to allow people to do that with their personal information.
Major Major Major Major
@PsiFighter37: people assume he’s a tech guy because all he does is talk about his tech CEO friends and his only idea is UBI, which is an idea basically wholly-owned by silicon valley at this point.
Yang’s plan is an Attention Deficit Disorder.
To what end?
Major Major Major Major
@Baud: why not? I can’t add my name to a mailing list to be entered into a contest to get a beer with Elizabeth Warren?
Trustbusting for the win!
Because look what happens…
Major Major Major Major
@Baud: in order to make sure that the sites only recommend government-approved content.
@Major Major Major Major: It would not take much digging at all to see that the dude has no relevant professional experience to back up all the talk about AI and technology that he is known for. His most meaningful experience (i.e. where he probably made most his money) was his initial starting job – a corporate lawyer – and whatever he made from Manhattan Prep, which is a test-prep company that catered mostly to large banks and consulting firms. Everything else reeks of being a ‘serial’ entrepreneur who never struck upon a big idea.
I have been very, very unimpressed with Yang since he first popped on the political scene, and his presence only proves that there are a lot of gullible people on the left who can be parted from their money as well.
@Major Major Major Major:
If it’s truly your property you can give it up or lose it, just like any other property. You want to set up an inalienable right. Just like you can’t sell yourself into slavery, you shouldn’t be able to lose your personal information.
@Major Major Major Major:
But what would the government be approving or denying? What’s the standard he wants to see applied to algorithms?
@Baud: I’m feeling a science fiction blockbuster!
Adam L Silverman
@chopper: Also, the Department of Redundancy Department!
Major Major Major Major
@Baud: so to be clear I couldn’t join a mailing list under your proposal? I feel like I’m missing something
Major Major Major Major
@Baud: His site just says
So whatever the government decides misinformation is, I guess.
Controlling social media algorithms will be like trying to regulate downforce out of an F1 car. The manufacturers just find some other way to achieve the same end while skirting all the rules you created.
@Major Major Major Major:
The simplest idea is that you couldn’t sell your identity to someone else to use. If it were your property, you could.
Major Major Major Major
@Baud: gotcha. His top bullet point there uses dumb language; the proposal in detail says “Data generated by each individual needs to be owned by them.”
@Major Major Major Major:
I’m not techie enough to know whether that’s even possible. Do algorithms currently used actually examine content for false or hateful content? (I know they try to screen for copyrighted material.)
@Baud: I alienate people all the time; why not people too?*
*Yes, I understand the actual concept.
This x 10¹² *! The United States understood by the end of the 19th century that too big is too dangerous, and put in place antitrust laws to control it. The only weakness is that we have to keep applying it over and over. We broke up AT&T, but is has reassembled itself terminator-like.
* Trying out the new formatting tools, plus a h/t to your nym.
@Major Major Major Major: Isn’t this more of a content moderation problem anyway?
@Major Major Major Major:
Also, whatever the government decides is “polarize” and “incite”. I can smell the freedom from here!
Major Major Major Major
@Baud: there are processes in place to pre-flag abusive content. For a funny example, a german twitter user recently had their account flagged for tweeting “die Boomer”, which is of course german for “the boomer.”
companies do less defensible screenings too, for example video-sharing site TikTok has recently decided to slow-walk approval for any video with a description that mentions Xi Jinping.
The “first wave” of content moderation is automated for many large platforms, and things flagged in that process are then reviewed by humans. For example, if you mention Chairman Xi or eating billionaires, you may get automatically flagged.
@Major Major Major Major:
Haha. Insert The Bart, The clip here.
The recommendation algorithms are not really the problem. They are based on what is trending, that is what is getting likes, shares, and comments, filtered by various demographic indicators for things that are similar to what you have liked before.
The real problem is targeting ads, and in particular the acceptance and targeting of political ads, especially if they are being purchased by dubious groups such as foreign government agencies, such as you know, the Russians. It should not be okay for Facebook to accept an ad from some Russian group advocating treasonous rebellion against the US government, and we need some administrative mechanism for regulating and preventing that.
@Spanky: I think at some point we need to have a rebalancing of the way the first amendment can be weaponized and cause real harm and the idea of absolute protection for it. Allowing the antivaxxer movement to spread debunked lies about the efficacy of vaccines is killing people, the ability for Russians and Republicans to lie constantly about everything is putting democracy at risk, and the ability for oil companies to lie about climate change is putting the entire species at risk of extinction.
Maybe we need to be a little less concerned about whether a totalitarian in power will start inappropriately restricting speech because we’ve opened the door to it (like they wouldn’t anyway) and be a little more concerned with keeping people alive and preserving our democracy.
I have a more micro-level question. When Facebook came on the scene, there were three proposals for social media to be more “portable”. These were:
Facebook rejected all of these and went for a winner-takes-all model. This is one of the main technical reasons why it is such a toxic swirl and people cannot get out of it.
Apart from FB being too big, a lower-level problem is that social media does not lend itself to competition due to “network effects” as they say – and that was supposed to be a more low-hanging fruit that could be solved.
If people could port their social graphs in some fashion, they can get free of the algorithms! Why is this not explored more? Why can’t this be mandated by a government?
@Ramiah Ariya: I’m not actually sure it solves anything at this point. The ad network can be taken cross service too.
There’s maybe more competition there, and the wealth generated by the data collection is spread across more companies, but it’s even more gameable by bad actors, as they can take refuge in the service with the worst moderation and spread to other services from there, and meme/post propagation isn’t going to be altered much.
All of the services combine into something that looks like what we have now, just with more complexity in the back end.
Major Major Major Major
@Ramiah Ariya: I like all three of those! Mandating some level of 2 & 3 is certainly worth exploring, probably as part of a data-ownership model.
I live in India. Politics here, before the advent of social media was not easily split into conservative-liberal, right-left dichotomy that exists in the US. But for the past several years I have noticed that FB and Twitter have caused this kind of a dichotomy to emerge purely in social media – that is, we now have an entire simulation of “What if our politics looked like US politics?” running across Indian social media – something that actually does not reflect reality at all.
I have assumed so far that this is because many people are influenced by what happens in America – BUT, it is also possible that purely American companies have created platforms now that somehow structurally reflect American politics all over the world. What would a social media created by a different culture look like? Nobody knows. May be it will present options or algorithms that are more locally appropriate? The exploring of this idea is impossible because of the gigantic nature of American companies and network effects.
This is whistling past the graveyard a bit. Warren is focusing on breaking these up, but that misses the point. Social networks only function because of their scale. The reason its so hard to break in as a new social network is that building a social graph requires scale – you can’t get your friends in your graph until they sign up which is why all of the newer networks are more broadcast focused rather than social sharing. TikTok, like Vine, is more of a lightweight YouTube than a video Facebook.
The analogue to true social networks is the USPS. It was invented as a way for any person to communicate with any other person. You got an address assigned by the USPS that was universal. And that was key. For some time, the USPS had a monopoly on the delivery of messages, but later UPS and FedEx and others got approval to do that. The USPS maintains the monopoly on your address – you don’t have a FedEx address, they just don’t get a monopoly on the delivery. But your social graph is maintained – my xmas card list is what it is and I can choose to mail letters or send packages via UPS or FedEx. If I don’t like how UPS handles my content, I can swap them out. but the ubiquity of the USPS address makes that easy. It’s why Amazon could roll their own delivery service so rapidly.
And this is the real failure of ‘small government’ – delegating the job of infrastructure building to others. And of course tech companies are picking up that mantle. Social media owns a big part of the social graph – Apples iMessage is also a very big asset in that space, probably second only to Facebook. Google has built out the social graph for work documents in GDrive, DropBox is in there as well for file sharing.
Apple has pretty much established how contactless payments will work, and nobody is really challenging them there. The biggest potential threat though is Apple which is building out the only truly secure identity system around the mechanics of why ApplePay is so dominant. They’re expanding into student IDs, looking at electronic drivers licenses and other identity forms – medical cards, etc. Unlike the previous infrastructure, which is mostly just regulating a users endpoint (which would also have the benefit of cutting down on fake accounts as well – the USPS doesn’t provide Russian trolls with postal addresses) Apple’s ID system involves on-device authentication, one-time keys, resolution of one-time encrypted messages over the network, remote identity issuance and resolution, and a few other bits and pieces. Its’ the kind of solution that’s VERY difficult to break up because it relies on so many components to work, many of which it appears only Apple has the technology to implement.
Major Major Major Major
I disagree. Think of America’s polling process: it’s far less game-able by bad actors because it’s decentralized across a bunch of jurisdictions, even if they’re operating from the same data concept (“US citizen”), even if some of them are way worse than others.
Extra complexity and competing integrations make it extra hard to manipulate things at scale.
Amen. I have a whole shitload of ideas for fixing the web. Out of all the problems with the web, autoplaying videos i a strong #3 behind progressive content loading (aka scroll jumping) and popup/over ads (particularly on phones where places like TPM can hardly be bothered to make sure these ads can even be closed on small screens)
But making it about kids is too narrow…the push for ad revenue and easy-to-use-but-hard-to-master tools like AJAX and JScript has made the web bad for everybody. (I categorize the whole thing as a shift in development practices from enterprise-quality to consumer-quality, aka buggy-as-fuck-all with shiny new baubles tacked on)
Major Major Major Major
So? Companies don’t have a god-given right to be so big that our options are “do nothing and let them ruin democracy” or “institute weird-ass restrictions.” Life was fine when I had to get my Harry Potter fanfiction on one website and then, horror of horrors, type in another URL to argue about video games.
There are articles on why social media graph porting is not as easy compared to mobile network porting. One of them is here. There is a robust debate on it, but it is a critical piece of the puzzle.
@Martin: I retired recently but the guy who was my boss used “scale” all the time and I never really knew what he was talking about?
Mo’ people, mo’ money.
@Eolirin: I get immediately uncomfortable whenever people talk about allowing or not allowing certain types of speech.
Fascinating. I’d never considered that possibility.
@steve g: No, the recommendation algorithms are hugely problematic when combined with bad actor content the algorithms are reliant on.
It’s now well documented that if you start looking at slightly right-wing content on YouTube you’ll eventually get nothing but InfoWars and white supremacist content in your recommendations. The way the recommendation engines work, as well as the mechanics whereby they queue up the next item without your direct input is a distilling process for whatever the dominant theme the recommendation engine finds. If those engines are tuned in any way toward left/right wing content and then weighted by new content, then they’ll inevitably take you to the extremes.
We are accustomed to systems reverting to the mean, but recommendation engines are intentionally designed to do the opposite. They are tuned to take you to the edges. Exposing new content is fine so long as you aren’t allowing new content to be dominated by propaganda, false information, etc. and it’s the combination of a first amendment with the ability for *any* kind of speech to get purchase and amplified which is the problem. We’ve never had anything like that. If you had extreme views, you’d get shouted down in the town square. Now, the people that would shout you down never even see you. In short, all content finds a favorable audience, no matter how bad/odious it might be.
Major Major Major Major
Yep. As I say in the post, even Facebook and Twitter (and Google, which has made their YouTube recommendation algorithm less odious recently) recognize this.
@Major Major Major Major: I’m not suggesting that they do.
But I am suggesting that there’s an underlying need which must be fulfilled in order for the breakup to achieve the desired goal. You actually give a good example of that – URLs are regulated. Very loosely, but regulated all the same. I can’t buy a .gov domain and pretend to be the US State Department in any believable way.
AOL used to have that kind of a private system where they provided the address of content. We replaced that single corporation approach (and they were comparably huge in their heyday) with a universal system that allowed anyone to serve as a directory. Ultimately Google Search filled most of that role, but it’s not like we need to take away the task of addressing content to replace Google (there are other things that secure their dominance in that space).
Anyone can build on this. But Facebook owns the addressing. So does Apple for iMessage, and Twitter and so on. That cant really be solved until the government in some form takes a regulatory role of opening those endpoints in a way that competition can come in. That’s got to be the first step.
I was working in local politics for a while in India, and I can tell you it is difficult to work without social media like FB, particularly its Groups and Pages. But there are huge areas for improvement in these, where, if competition can exist, we will all move to. Let me give you an example:
FB Groups has a Owner and can have multiple Admins. When you have a political party’s local branch having an FB Group, it differs in structure from the ACTUAL branch. A local branch of a party has a committee which makes decisions and acts on them. Nobody can cease power over the branch by themselves. But FB Groups do not allow this kind of a community decision-making structure in either content moderation or administration. Instead it is possible for one of the Admins to cease control and boot out everyone else (this has happened). This is a huge problem for any political party or organization, and if we all could collaborate using a better social media platform we would.
Major Major Major Major
@Martin: This is a proposal I could definitely get behind, thanks for elaborating.
Huh? It’s just a payment system and NFC. Any vendor that accepts Apple Pay can almost automatically accept other payment systems as well. And didn’t Apple recently overtake Starbucks, which previously was Number 1 for these types of payments?
It’s important to recognize what happened at the end of the 19th century. The precipitating act on the railroads wasn’t that they were too big, it’s that they owned the rails. If you wanted to get from A to B, and Santa Fe owned the rail between those points, you had no choice but to use their service, or as a competing railroad to lease access.
What changed wasn’t breaking up the railroads. It was the interstate commerce act which opened up access to those rails by regulating them, regulating the prices they could charge, and so on. It was the exclusivity to their network that was the problem, not their size. It’s generally the same problem now.
Another problem area is mobile carriers. If the FCC regulated the carrier frequencies and protocols and separated the ownership of the towers from the endpoint services, then we wouldn’t have problems like 4 mutually exclusive wireless services in big cities, and none in rural areas. Only recently has the technology advanced to allow cell phone makers to include radios that could work on all networks, so you didn’t need to sell your expensive phone simply to change carriers.
Regulate the infrastructure and allow competition on the service.
@Major Major Major Major: Don’t underestimate the power of the ID part. Apple is consolidating power there in a way that almost nobody seems to be aware of.
Very interesting question. I’ve read that China uses a different social media platform and also rigorously controls it. However, the government’s level of control also makes it difficult to analyze the impact of the platform.
But I would also think that how societies and sub-cultures choose to use platforms might also have some influence. Years ago, in Brazil, a google social media platform was dominant and seemed to serve its users well. I think it was called Orkut. Google tried to move these people to google plus, which was a massive failure. I guess they all ended up on Facebook.
But I guess we are still talking about American companies. Haven’t some other Asian social media platforms become popular in the West?
You could post all the algorithms you want and I wouldn’t understand any of them.
I kind of feel like whatever the government mandates will have a workaround by the next business week. China is trying to get face ID on everyone and grade people by social media activity, but the Hong Kong protesters have used masks, umbrellas, and pulling down camera towers to avoid being identified.
Major Major Major Major
Well, we know what Chinese social media looks like. And many regions have their own versions of niche media, for example, the gay app they use in Japan is super Japanese.
Yes, the social credit system seems extremely insidious in many ways. From the Wiki:
And I think that sometimes a person’s social score is made public.
I’m not sure that even authors of dystopian novels quite saw this coming in the way it has.
@Major Major Major Major: We’re talking about a system of interoperable social network services that can communicate back and forth with each other? If one of them shuts down an attack vector, you can move to another that doesn’t and get in the back door, because you can have that content linked to without running afoul of moderation.
I view this as more akin to what would happen if you allowed health insurance to be sold across state lines.
The target isn’t the social media networks, so you don’t get resiliency from making the social network systems themselves more robust. Since you’re just targeting people, broadly, you just need one weak entry point to propagate and spread from especially since you want to be able to mainstream it via more official channels anyway.
Might also have some relation that so many think that the American system is the absolute best system. But other systems exist and work, some better, some worse, probably depending on your point of view. And our system hasn’t evolved well when you consider things like guns or voting systems or who we sometimes elect as leaders.
We all are flawed human beings. Some with more flaws than others, some with more flaws than many others combined…. Give those people power and all hell breaks loose, as we can see in many parts of the world today.
Very interesting insight. And the algorithms may not discern curiosity from strong partisanship. I once clicked on a “cute parent interaction” video simply because it had been widely publicized and was in fact very cute. Now I get served others despite my general lack of interest.
I also wonder if we really understand the degree to which bots and deliberate efforts to direct and subvert political conversation has been effective.
But the most malicious counter-reaction may have been the very success of the Internet. It became so easy to find facts and true information that people who believe lies, myths, tradition, ideology, etc., felt a need to create sites devoted to their own misinformation and to do whatever they can to spread their beliefs.
The ultimate representation of this is Trump and his Twitter rants. This in turn helped feed the lie by conservatives that Google and other sites were being unfair to them by not ranking their bullshit more highly.
I’m finding it increasingly difficult to believe that Andrew Yang is just a young, well-meaning guy. I cannot understand who his actual or desired constituency is. I just dont see any evidence of an understanding of tripartite system of government but more of a Springfield Monorail
One only has to look at our congress to see the lowest common denominator and the freedom of it swinging the overall direction.
Is this the cost of freedom of speech and thought?
Or is the cost of repression that we don’t even see the other side and have some chance to decide for ourselves?
We see the first point here, right now it’s called the conservative side of the issue, the response to freedom is the call for repression.
And that brings the second point that repression shuts down even the concept of difference.
Which is better?
I’m going with the freedom, because it allows a difference and that is better.
@Brachiator: Actually, it’s not. It’s extraordinarily clever, actually.
For ApplePay you start by establishing a secure identity connection with your card provider (the banks decide how they want to handle that, which is almost universally not-great, but some are getting with the program) whereby they issue a secret account ID to your phone, which is stored in a secure part of the hardware.
You cannot see that ID, and the operating system has no access to it. It’s unique to your phone and only valid on your phone. If you replace your phone, that ID needs to be reissued to your new phone. There’s a whole layer for identity management around this, allowing for trust chains and the like, so for example, if you had a similar secure ID for your SSN, your bank could ask to validate your secret SSN ID with the Social Security Administration as part of their authentication process for your credit card. It’s not really being used in this way yet, but it’s there. Apple does use it for authenticating your Apple ID though, but I think that’s the only chain of trust application out there. Maybe the student IDs use it.
Anyway, when you go to make a purchase, the phone creates a one-time card number using your ID and a unique identifier in your phone (again, which you don’t have). This one time card number (a token) can’t generally be resolved into a card number. Only a token resolver, who is an entity along the transaction pipeline, usually at the very end, can do that. To the merchant it looks like a credit card number, though one that won’t resolve to industry numbers (your Visa won’t necessarily start with a 4) and one that deliberately won’t checksum using the last digit. If they store it, it won’t do any good because it can’t be used after the transaction is done. Therefore they can’t leak it in a damaging way.
The token resolver has keys from Apple that decodes the number into a bank recognizable account before passing the information along. I don’t know for sure, but I think at this point it may go through a separate re-encryption just between the token resolver and the resolution (where the ‘funds’ are transferred from your bank to the merchant’s bank). The reason I think that may happen is that’s how ApplePay got started. Apple worked that arrangement out with the iTunes Music Store/App Store years ago, so Apple never actually stored your credit card on their servers (which is why Apple, with a billion credit cards on file, never leaked – there was nothing to leak). Instead they created this secret ID, ran the one-time encryption, and had it decrypted by their bank so that the transaction couldn’t be intercepted.
The system is clever because it’s almost impossible to break. You as the user never have access to the identity information, so you can’t be phished. The merchant never has the information, so they can’t leak it. If anyone did get it, they couldn’t use it without physical access to your phone – information which Apple bakes into the CPU when the CPU is made (part of why Apple’s system is so good is they make their own CPUs and therefore can ensure this is happening correctly). And Apple helped set up the system of token resolvers to intercept these and resolve them appropriately.
The reason why it’s going to dominate is that the ~1.5% overhead for fraud doesn’t need to be collected for ApplePay – but still does for most other contactless systems in the US. With chip+pin that could be reduced, with chip+signature, it’s still mostly there, and most other contactless systems are just mimicking a chip+signature transaction, which is a bit more secure, but still vulnerable. For instance, it doesn’t solve card on file transactions (Amazon also uses Apple’s system – that’s what they got in exchange for Apple’s getting their 1-click patent, but Target, Walmart, etc don’t). Or for online transactions where you type in your card number.
ApplePay also has a mechanism to do secure online transactions using the same system. The key is the ability to use your local authentication (FaceID, TouchID) to unlock the identity on your physical device. But it can only work in Apple’s browser because there’s no way to open up the entire process end-to-end to 3rd parties.
That’s why I think it’s a difficult problem to solve in a regulatory way as it requires regulating everything from how the CPU is made to what access the OS has to that information, to the encryption systems, to the transport protocols and so on. It’s not impossible, but so many parts of it exist only inside Apple that I can’t envision a scenario where it could be replicated on Android.
I’m going to go with – himself. Or possibly himself and his money.
@Mary G: It’s actually a bit worse than that.
Traditional programming takes data and rules and creates solutions. Machine learning takes data and solutions and creates rules. The resulting rules from any ML tool, which are pervasive in social media recommendation systems aren’t algorithms as we normally experience them. They’re largely statistical weightings to an array of variables – sometimes a MASSIVE array of variables. Understanding how they interact is nearly impossible to discern. After all, if we knew how to get to that outcome we could have done that with a traditional algorithm, but we couldn’t get there.
It then sounds temping to simply ban ML use in recommendation engines, but invariably that will be the solution, even if it means doing it from outside the network. You can slow it down, but not stop it. Better to go to the root of the problem rather than kick the can down the road, IMO.
Right. What the studies are finding is that the algorithms are able to turn curiosity into strong partisanship. In effect, it’s able to transform individuals from being economically anxious into white nationalists.
We think we have a lot of free will on this stuff, but we really don’t. This is a really important video to internalize. It’s focused on economic decision making, but it would apply equally to shaping political views.
@Martin: “We are accustomed to systems reverting to the mean, but recommendation engines are intentionally designed to do the opposite” I was with you up until the “intentionally designed” part, and your idea that the sites look for “themes” in the content. As a techie who has worked on the problem of recommendation logic before (though never on anything very sophisticated), I think it’s very implausible that anyone at Youtube or Facebook either set out to encourage extremity per se, or was even capable of creating an algorithm that actually understands the content in that way. What it looks like they’re doing instead is delegating the task of knowing what the themes are to the creators and viewers of the content– that is, people who post a lot of Nazi shit tend to also post more of the same, and they follow people who do the same. It would do the same for people who are super heavily into birdwatching; we wouldn’t consider that to be ideologically extreme, but it’s a cluster of people who are into the same stuff as each other and most other people are not into that stuff, so the algorithm is making a reasonable inference that if you are looking at some, you might want more. I don’t think this represents any kind of strategy of deliberately making people into bigger fans of bad shit (or birds). It’s a half-clever design for making a system work on a large scale with minimal human understanding required. But only half clever, because like any system that isn’t operating on actual understanding or even an actual preference but is just trying to imitate people via statistics, it can easily get into weird feedback loops that aren’t part of the intended purpose. For the same reason, I’m skeptical that Yang’s idea of making the algorithms visible is even possible in any useful sense. I’m pretty sure what we would see is a lot of stuff like “pick the most closely correlated 10 things out of distributed database A that were put there by services B, C, and D and were cross-referenced with database E by service F based on the last 10 million inputs we received,” with absolutely no reference to any specific type of content.
(ETA: I see from your later comment that you’re well aware of what ML is, which makes me even more puzzled as to why you were talking about intentions and themes.)
Major Major Major Major
Hong Kong comprises 0.5% of the population of China. That shit wouldn’t fly on the mainland, and won’t fly in HK forever. China is only not responding (very) violently because they don’t want the bad PR.
@Major Major Major Major: Is there pre-screening of content containing Winnie-the-Pooh, too? AFAF.
(Explanation of joke for those who don’t possess a bit of possibly esoteric knowledge; highlight to view: For those unaware, Xi gets really offended when people compare his appearance to that of Winnie-the-Pooh. I don’t even see the resemblance, but I find it hilarious that he gets so upset over it.)
Re: Yang, I want to like the guy because he’s basically right about automation, and universal basic income is the best solution I’ve seen to deal with it… but the dude isn’t qualified to be president, and when he talks about anything that isn’t UBI, things frequently go off the rails quickly. And sometimes when he is talking about UBI – that stunt at the latest debate wasn’t great. At least people are talking about the idea more, I guess.
Ceterum censeo factionem Republicanam esse delendam.
I’ve been resisting Apple Pay because it seemed like it could be just one more leak point. But I know that the phone picture identifying me is rather secure because I have to do it correctly or it makes me type in my code. IOW it is very selective for an ID. If the payment system is as tight, as impenetrable or even more so, then it really can be a rather secure payment system.
Can someone with more experience and knowledge of the tech industry explain why breaking up FACEBOOK and GOOGLE and AMAZON any other big tech companies would lead to a huge flowering of new jobs and innovation as the smaller pieces start innovating and competing and more space is opened up for new firms to compete?
I would think that breaking up FACEBOOK would be the best thing ever for software engineers and startups and everyone else in that world.
Likewise, it would probably make Zuckerberg richer if he own pieces of all the resulting independent companies.
What am I missing?
Major Major Major Major
Andrew Yang is the sort of person who’s been saying self-driving cars are five years away for the last twenty years.
@laura: No, he’s well meaning. But tech bros tend to not be social scientists, and we ignore social scientists at our peril.
He’s offering up tech-oriented solutions for social problems. Some of them completely miss the mark. Warren is a great candidate for technocratic economic solutions. Yang is offering up technocratic technical solutions. Harris is technocratic legal solutions.
They each have their wheelhouse, and they each miss outside of that. Harris is my preferred candidate but her solutions for housing affordability kinda don’t make sense. Warren would get that one right.
@Martin: The revolutionary system you describe is called PKI. First introduced in the 90s. Not fool proof but more secure than going without.
@Major Major Major Major: Hence my qualifier of “basically”, but I still should’ve been more specific: he’s basically right about the impact it will ultimately have on our economy, though he’s probably wrong about several of the specifics and maybe even the timeframe. I’m typing on a phone, so I didn’t feel like getting into particulars.
Ceterum censeo factionem Republicanam esse delendam.
Are you suggesting that Apple may become the behind the scenes provider of payment services? I think it unlikely, but I don’t have a problem with this.
I understand that historically, some merchants refused to use any contactless system. And banks and other merchant service providers fought it.
I use Android and Google Pay. I don’t have an iPhone and probably will never use an iPhone. But if Apple becomes the backbone, again, don’t much care.
Europe uses a chip system. They seem pretty happy with it. Somehow, I don’t think that the EU will allow Apple to become the sole provider of payment services.
Looking at this from another angle, I will be interested to see which company will provide the most secure card or device or system which allows for a near-universal cashless society.
Martin, that is quite an education on Apple Pay. You are one smart dude! Myself, I’ve been using Samsung Pay since shortly after it started, for about 3 years it seems like. I used to get a lot of “Apple Pay doesn’t work here” from clerks (what brand recognition!) in years 1 and 2, but lately not so much.
Apple Pay is indeed #1 (https://www.phonearena.com/news/apple-pay-most-popular-mobile-payment-service-us-starbucks-second-place_id119885). I had no idea Starbucks had a payment system.
Establishing property rights to ephemeral things reminds me of when Kramer sold his life stories to J Peterman for $750.
@Ruckus: There is nothing remotely as secure as ApplePay. What’s remarkable about it is how well they’ve plugged so many disparate holes – phishing, merchant leaks, skimming, etc.
That said, some parts of it really only work in theory. The new Apple Card is an attempt to close that. Banks still issue traditional card numbers which are easily phished, even to people using ApplePay. Apple Card is an attempt to close that. But merchants are slow to get rid of their legacy POS systems and hook into this end to end. My bank issues a credit card that works with ApplePay, but not a debit card. So it’s hard for me to use the Apple Cash money sending system which uses the Federal Reserve Clearinghouse on debit cards, which costs almost nothing, but does a cash transfer if all it has is a credit card, at a cost of 3% or so. Bank of America and some others are putting ApplePay into their ATMs so you can withdraw cash without your card+pin. So, you can’t skim that transaction. You do need an Apple device, though.
Another aspect of the system is that if your phone is lost, you don’t need to get a new card issued. Instead you just deactivate that key and get it reissued to the new device. It doesn’t even affect other devices you might have (like an iPad), they can keep working. The ability to isolate this on a per device level means that if you could get rid of the physical card itself (as Apple Card is attempting to do) then you’d never need to have a card reissued unless you changed banks. You’d just constantly deactivate/reactivate IDs attached to that card. One of the new features Apple is introducing is the ability to give every card on file merchant a unique card number, so again you wouldn’t need to change out all of your auto bill pays every time your card got upended.
This whole setup is amazingly good for consumers, but with the downside that the reason it works so well is that it shifts a LOT of the key efforts onto your physical device. That’s going to consolidate your financial records onto your phone, which puts it under Apple’s ability to control how it goes out, which will invariably lock out competitors.
In effect, Apple is building a nearly impossible to break identity platform centered around your physical device that can cross identity platforms. In theory your bank could securely verify your SSN without your input (also meaning, you lose your ability to lie to the bank), and cross check that with your DMV, etc.
This exact same system also undergirds Apple Health’s new VA record system (which you might personally benefit). Not sure to what degree the VA is bought into this, but the system would allow the VA to issue a secret ID to your iPhone or iPad or whatever after confirming your identity via other means, and then use that to retrieve and encrypt records between them and your device without having to touch Apple servers, and you can then securely send them health information, either collected from your devices like the Watch with the EKG function or other means.
If Apple can hook the VA into their systems, I see no reason why we couldn’t see SSA/HHS, state DMVs, etc. But the only thing unifying this identity system is ownership of a bit of Apple hardware.
I don’t think this is true at all, certainly not for the vast majority of people. When the algorithm gets things wrong, it just annoys me. But it never induces me to click on much stuff that does not interest me, especially when it comes to politics.
I agree with an inference drawn from your earlier point that the easy discoverability of extreme and fringe points of view makes it easier for extreme partisans to attract others and to reinforce one another. And your key insight is that these people are never shouted down or presented with alternative views. Or more precisely, they can easily exclude alternative views and feed on extreme views. You get confirmation bias magnified to the ultimate degree.
An interesting hypothesis that with respect to ideology the Internet resists regression to the mean.
I have to bounce now, but will look at the video you linked later on.
How does he want to solve the problems he sees?
And that’s not how to solve a particular problem it’s how to solve it when it involves humans, which of course it always does in politics.
Which is I believe what M^4’s answer was about. Humans can solve complex problems such as driving a car because most of them are able to see more and think about resolution and focus automatically, after training and experience. Electronics have to learn as well but don’t do that the same way and have limited storage ability to use the necessary responses at the necessary speed. Some humans don’t have that either. For example, I raced motorcycles and the focus and thought process to do that is a learned one. Some learn more and were faster than me. Others were slower because they learned and responded even slower than me. I was barely better than average. A machine can learn in that same method but they learn a more specific response, rather than generalization. And that makes the response more specific, which often is the wrong answer. The generalization is the point of that learning curve in racing, how to react to a specific problem in a generalized way, rather than with a simple fixed response. The situation is too complex to learn a specific response, as in always do this in response to this. One man wrote a book about how he learned specific responses to specific problems and he was a winner. Until others learned that a more generalized answer solved the problem faster. His specific response concept worked up to a limit. A generalized solution worked past that limit.
@Brachiator: No, they don’t want to be a payment processor. Instead, they’re taking 0.1% off of every transaction in exchange for eliminating the need for the bank to pay out about 1.5% for fraud. The cost of reissuing a physical card is between $5 and $15, and ApplePay eliminates a lot of that cost as well.
That’s the other clever part of this. They’re critical to the system working, but they effectively provide no services. The key bits are bought by the consumer. It’s as close to a 100% margin business as you can get. It’s not quite that good because Apple also implemented a system where they work with the issuing agencies (banks, etc.) on support calls, so Apple Support staff can see the same information the banks can, so they can seamlessly hand you off from one to the other if there’s a problem. But I see that as something with diminishing costs as this becomes more common.
It’s very lucrative to just be in that spot. Apple earns $1 for every $1000 you spend (vs the bank having to charge and pay out $15 for fraud issues). There were 3 billion ApplePay transactions last quarter. Average credit card transaction is $88, but I think ApplePay tends toward the low end. Buying coffee and shit like that. Let’s say it’s only $10. That means Apple made $30M just last quarter off of those transactions, while doing nothing but getting all of the pieces in place and letting others pay for them. Not bad. If average transaction is closer to $88, then Apple made about a quarter billion last quarter with roughly 100% margin.
So they get all the control of owning the processing system, without owning any of it, and the cost of entry to the consumer is buying an iPhone. It’s an arrangement that I know economists are struggling with, and one which I don’t think government regulators know what to do with.
Some merchants don’t understand that if their systems are set up for Apple Pay, they will work with other systems as well. I asked a store owner that had an Apple Pay sign on his door if he used Google Pay. He said he didn’t know. But I was able to use my Android phone no problem.
I had a weird situation not too long ago where my bank sent me a new credit card because of a suspected security problem. The scary thing was that I used this card at only a few places and at a terminal. A sales clerk was never in possession of my card. And yet, still there was some potential problem.
So far (fingers crossed) I have never had a similar problem when using Google Pay. Or even, to think about it, my Starbucks card.
@mad citizen: I’ve been researching it for a number of years. Currently sending my kids through college on that information.
@Brachiator: Yeah, Apple had to thread a very difficult needle to this to work. What was key was finding a way to do it that wouldn’t break other system, because as a new entrant they knew they needed to piggyback on existing systems. The upshot of that is that you get virtually all other contactless systems along for the ride, so any ApplePay processor should take Samsung/Android Pay with no problem, even if they don’t realize it.
Major Major Major Major
@Ruckus: sorta. Automatable tasks will probably eventually be automated, but Yang’s definitions of “automatable” and “eventually” are very different from mine. As you note, discrete tasks, such as playing go, or even basic translation, are where we should *expect* AIs to shine, as indeed they do. But as it turns out most things are not playing go.
I understand and agree.
There is always a cost to any system/idea. Apple has solved, at least far better than anyone else, a system for money usage in a modern world. And anyone with an Apple product pays for that. Including me. Is the cost acceptable? Well I use their computers/software, iPhone because they work better in my mind even though they cost more. The increase really isn’t that much. Two computers ago I did a cost comparison for the total system – hardware/software. The Apple system I bought cost less than $50 difference on point for point. I could purchase a non Apple product for less/higher difference but it lacked in many points of comparison. Apple doesn’t seem to sell loss leaders just for the market share. In my mind as a two time small business owner, a good business concept.
I don’t see masses of people buying an iPhone just to use the payment system. Android and Samsung provide similar services and “good enough” will work. And developing companies are doing quite well with payment systems tied to various devices. What might be interesting is a potential disruption of banking systems linked to devices and cards. In the US, some lower income people are excluded from having savings or checking accounts because of banking industry bias and greed. You also have people having to resort to payday loan services. But in the tax prep industry, for example, people can have their refunds loaded onto a card which they can use at selected banks even if they do not otherwise have a banking relationship. I will bet that a lot of lower income people have smart phones even if they don’t have a bank account. And this creates a huge opportunity for Apple and other companies, should they choose to get into this market.
ETA: I see Amazon exploiting new payment systems more, not just for their main site but for Whole Foods. They may already be moving in this direction.
Major Major Major Major
@Brachiator: don’t worry, Facebook is releasing a cryptocurrency for them!
@bemused senior: Apple’s contribution is that they take seriously both software and hardware (physical) seriously. That is, in important ways Apple is a computer security company, which sells oh-so-shiny devices that command a premium price based mostly on their shininess. The “Secure Enclave” is a big deal. Here’s some ancillary testing work, where certain attacks via changes to device physics (there are others) are evaluated. (As a side effect, but still.) Inside Apple’s top secret testing facilities where iPhone defences are forged in temperatures of -40C – UK exclusive: In a never-before-seen look at its privacy work, Andrew Griffin assesses criticism that the tech giant’s data security techniques limit the benefits of its products to the rich (Andrew Griffin, 27 May 2019)
DO you remember the Target credit card hack/debacle of not long ago? My small 5 star bank stopped transactions of all of their issued cards. None of the big banks did anything at all. I had to wait about 2 weeks for a new card, and we know that pretty much everything is a no cash transaction any more. But I didn’t lose a dime and as best as I can tell, neither did the bank, other than the cost of card replacement.
If Apple Pay cuts out that issue it’s a lot better than waiting 2 weeks for a new card, which is no more secure than the last one, it’s worth the price of admission. At least to me.
@Major Major Major Major:
We humans have our faults. Oh boy do we have faults. But what we do have is mostly the ability to actually think about things like driving, a complex issue, and make it work for the majority of people, a majority of the time. I doubt that machines will get there, only because they have to think inline. That’s not the right term but the concept is there. Yes we have far faster thinking machines and multiple thought channels within them but they still have to think in a linear fashion. Humans don’t have to do that, when things are working well we can think about the world we see in a far different fashion. That methodology has flaws for sure and as we age or if we have brain issues that ability may/will be as flawed and not equal but it still gives us the opportunity that I don’t see machines having for a long time.
I’ve been involved with electronic machines for 46 yrs and while the process has gotten so very much extremely better than the original very basic machines, they still have limits based upon linear process.
J R in WV
Really? you really don’t think mailing lists aren’t for sale to any one with the $$ to pay for the mailing list, by zip code?
I think the RNC bought all the addresses in my Zip code, which is how the Trump campaign could send me a pack of lies while seeking financial support from me, a life long Democratic voter!
Major Major Major Major
@J R in WV: oh yeah you can buy anybody’s address.
So, haven’t thought a lot about the policy implications, but I think I’m basically OK with recommendation algorithms, as long as the links don’t send me to dangerous or bogus sites that lie to me or rip me off.
Contrast with TV – I try to DVR everything, but when the Seahawks are playing, I like to follow along in real time. So now that we’re in the “holiday season” we’re bombarded with ads from companies like Lexis that have these stupid Christmas morning celebrations in the driveway of some fake family’s airbrushed home, as they ooh and ahh over a new car with a stupid ribbon on top of it. Now I know some people buy these cars, and some subset probably wait until a festive moment to debut those new purchases. That’s fine, but even if I had the money to do so, that wouldn’t be me.
So back to the internet, any human being, and most bots, could pretty quickly ascertain that I’m not a holiday Lexus buyer, and not worried about accidentally eating the vegetable that Gundry MD assures me will gunk up my guts. I’d much rather see ads for grateful dead t-shirts, fly-fishing gear, or single-malt scotch. Maybe if I get an unexpected windfall, I’d be interested in a new Volt. But never a 4×4.
Isn’t there a non-evil way to do this channeling? Maybe give the Consumer’s Union regulatory authority, paid for by a tiny transaction charge?
J R in WV
Will never work for us, we will never own a “bit of Apple hardware!”
Just never going to happen, have used Apple tools when visiting Apple user friends, was counter-intuitive, impossible for a non-Apple user to figure out without a long interval of experimental try and fail time before figuring out how Apple thinks intuitive works.
And I have 30 years of experience from multiple IBM mainframe OSes, CP/M, DOS, Windows, Unix, Linux… but Apple toolsets are opaque to me. Hates them!
@Brachiator: Apple has 50%-60% of US adults. That’s a lot. And I agree just for ApplePay people aren’t going to switch. But right now there are parts of the VA system that are more integrated inside Apple Health than anywhere else. If those things start moving forward solely on Apple’s platforms, that starts to be a real challenge.
Facebook has that kind of ubiquity in a number of foreign markets. Many businesses in India don’t even bother with websites instead opting to go with a social media storefront. That’s a little troubling to me. Apple at least is threatening to have that kind of potential impact in terms of access to consumers.
@Ruckus: He’s proposing universal basic income as a solution to tide people over during periods of unemployment. This seems almost inevitable, as automation will cause people’s employment status to become increasingly unpredictable and beyond their control.
My career trajectory may not be a completely reliable barometer here, but I’ll lay out some basic details anyway. My first employment was in retail. I came to hate it after a long enough period of time, and as I didn’t see any opportunities for advancement (being autistic, I do not possess the requisite skill set), I eventually quit.
My second employer was a well-known TV ratings company. I won’t mention their name in large part because you’ve already thought of it. I began working in a position entering people’s TV viewing that they’d written out in diaries into a computer system. I was good enough at this task that I was promoted within about three months to another position that involved ensuring that the listings of TV stations, cable companies, and other similar data in the computer database were correct. I did this for about two years and became quite good at it, but the entire TV diary program was shut down in 2018 after the entire process of recording TV viewing in smaller markets was automated.
My current work is for a company that does audio transcription. Due to confidentiality laws, I am legally forbidden from disclosing any specifics of my work; I think it’s probably safe to say that the large majority of jobs we get pertain to finance, which naturally involves a lot of confidential information. My original position involved simply transcribing the audio. After approximately a year, I received a promotion to a quality assurance position; I now spend about 50% of my time doing quality assurance with other people’s jobs, trying to clarify portions they might not have gotten or correct mistakes in their transcriptions. The remaining 50% is still completing my own transcriptions.
My employment status is relatively secure for now, but I’m fully aware that this is another task that will probably be performed by machines within my lifetime. There are already attempts to program AI for this task, but for the time being, it’s not reliable enough for our clientele.
I do see a number of potential cases where existing software could improve our efficiency. A number of our dictations have awful audio quality (which is the reason for a large number of QA jobs I get in my current position) and could be cleaned up to the point of much better intelligibility with noise reduction software. I presume we don’t do this because it would be a complicated, technical task that would slow down our turnaround time (a lot of these jobs are time-sensitive, and we try to get them out within four hours during normal business hours), but I suspect it would already be possible to program an algorithm to automatically detect noise within a signal and apply noise reduction in cases where it would help. This would in fact improve our turnaround time and accuracy if it were done correctly. However, it would compromise the security of my own QA position, as signal noise is the primary cause of many of the QA jobs I get.
There are cases like this throughout our workforce, where technological improvements can improve efficiency and output, but compromise individual labourers’ job security. One proposal is to re-educate workers who lose their jobs to automation, and there is a strong case to be made for this, but it is not an instant solution, and it’s perhaps not possible for everyone to, for instance, be taught to code.
Moreover, in many cases, the jobs brought about by new technology may not coincide with the places where people lose their jobs. The decline of coal has coincided with a rise in the number of green jobs. However, by and large, those green jobs are not where coal miners live.
We could perhaps pay people to relocate, but not everyone may want to. I don’t want to leave Florida right now because I have too many connections to the area. My family and closest friends mostly live here; most of my support system is here. If my next job offering required me to relocate to L.A. or D.C., I’d probably be thrilled in some respects, but I don’t know if I’d be able to get by in either location, given how expensive they are. (Again, I’m autistic. Well, “on the autism spectrum”, but I hate that term – it’s psychological jargon that most people can’t comprehend, and it obscures the severity of the obstacles I face.)
So… what’s the actual solution? Basic income is the best one I’ve seen. Just give people just enough to get by during periods of unemployment, but if they want to purchase luxuries, they’ll need jobs. The biggest objection people have, “people will just stop working”, seems to be largely baseless. The Mincome program in rural Manitoba operated along roughly the lines of the program Yang is proposing. IIRC, the only people who stopped working in any large numbers for any significant periods of time were new parents and students.
It turns out that most people like working because people want to feel productive, and they want more than just the basic necessities of life. People like going out to movies. They like going out for dinner. They like attending sports events or purchasing music or watching cable television or whatever else. Food, water, and shelter aren’t enough. They’re necessary but not sufficient.
Basic income is the best solution I’ve seen to counteract the increasing unpredictability of employment in the coming future. I’ve also seen a suggestion that, alongside that, we limit the maximum amount of hours people may work. I’m not sure how doable that is on constitutional grounds (this is something our lolyer types will have to weigh in on), but it may be something worth considering. Americans, on the whole, work way too much. 40 hours a week is pushing it – it’s the maximum people can work over a long term without a loss in productivity. 50-hour sprints can be manageable for a month or so, but after that, you lose productivity, and you’ll probably need time to recover afterwards (so 30-hour weeks).
But I can see a case that just because we can work that long doesn’t mean we should. Maybe limiting people to 35 or even 30-hour work weeks is something we should discuss. And maybe when enough jobs are lost to automation, it might even be worth discussing a 20-hour week.
Yang is completely off base with a number of his policy proposals, but as far as I see, he’s the only candidate discussing how much automation is going to change the way we work in the future. There’s a lot he gets wrong even here. But in some respects, he’s running for an election in 2020 on a platform that I suspect would have much more relevance in 2040. I don’t endorse his candidacy, and a lot of his fans are complete whack jobs, but I am at least glad he’s inducing people to talk about automation more. In that respect, at least, he’s a much better candidate than someone like Williamson or Gabbard, who are respectively a complete crank and a probable Russian asset. I may also prefer him to any of the interchangeable mediocre white guys who may have more impressive political résumés on paper, but are not making any novel contributions to the debate and have made no convincing case for why they should be president or why they can defeat Donald Trump.
But Yang hasn’t really made any case for why he should be president, either. His slogan about how the opposite of Trump is an Asian guy who likes math is kind of funny, but it’s not actually an argument for why he should be president. But I get the sense that he’s not really running for president. He’s running for the same reason Inslee ran: to increase discussion about a specific topic. If that’s his actual motivation, then I think I’m fine with him running. But if he actually wants to be president, then he’s just as much of a crank as any of the mediocre white guys.
He should run for some other office than president if he really wants to craft policy, though. And that’s part of the problem: all these people who run for president rather than trying to build political résumés from lower positions first. At least Inslee and many of the mediocre white guys have been governors and senators. (If you aren’t at least a governor or a senator, you should run for one of those offices first – and I include Buttigieg in that category, though in his defence, I think he first started out his campaign as a run for Vice President or a cabinet position, and was just as surprised as anyone when he became a serious contender for the nomination.)
I dunno. Obviously, I’m very conflicted about Yang. I at least like that he’s sparking discussions about automation and basic income, because automation is already causing seismic shifts in people’s employment status, even if our technologically illiterate media don’t understand it yet. This will only increase as technology continues to advance. In this respect, he’s really running for president of the United States of 2040. He’s gotten a number of specifics wrong, and he’s completely off-base on a number of topics that don’t pertain to automation or basic income but at least he’s introducing a few new ideas into the national discourse, so I’d rank him above the likes of Tim Ryan or Seth Moulton or John Delaney or any of the other completely mediocre white guys I’ve already forgotten.
Ceterum censeo factionem Republicanam esse delendam.
@J R in WV: you’re confusing things. Yes, you can buy a mailing list. That’s equivalent to assembling a social graph. But a mailing list isn’t an endpoint. That’s what I’m getting at.
There’s an economic cost to getting an endpoint (a postal address) that prevents setting up massive bot accounts because those messages have to also come *from* a valid endpoint. It’s a federal offense to not do that. That system puts enough economic cost on creating fake addresses and enough penalties on abusing the system to keep it reasonably in check. The reason we get frustrated at politics is that political calls/mail tend to have exceptions to those policies which opens them to abuse (which dudes like Roger Stone always benefitted from).
The graph and delivery can be more open and less regulated because the endpoints are more secure, and you can take a given address and resolve it to an actual person responsible for that address. You can’t do that with social media. Most addresses can’t be resolved to a person. Facebook has deleted over 3 billion fake accounts. There’s only 7.5 billion people, so roughly one account in two is unaccountable to a person. That’s the real problem. Tighten that up so that when I see a message I know it’s from a dude in Russia and not a fake dude in Ohio, and a lot of that will die down.
@Ruckus: The problem isn’t that self driving cars is harder. We can currently build autonomous cars that are less likely to cause accidents than human driven cars. So, statistically, it’s a solved problem. What’s more, there are networks effects here. The more autonomous cars, the less accidents, because their behavior is more predictable.
The two challenges with autonomy are:
The other problem is that we have some shitty autonomy efforts. Ubers was atrocious. Those cars should never have gotten on the roads with the software they had. Tesla is better, but they’re really pushing the envelope more than they should be. Waymo (Google) is actually very good and deserves tremendous credit. They have their shit together. And though they’re safer than human vehicles, they’re routinely taking a long view. Another year of development to unlock a $2T market is safer than rolling out there like Uber did and running people down, destroying your effort, in a rush to be first.
Autonomous vehicles are ready now. My contention has always been that drayage trucking and then long haul would be the early markets. Those have much simpler routing issues. They only use maybe 5% of the nations roads, and they tend to be well maintained roads. What they might need to upgrade would be pretty straightforward to do. And the economic argument is there as well. Drayage can adapt to an electric/fuel cell drive train and autonomy since it tends to be from a single fixed endpoint to a limited number of destinations (port to distribution). Not everything, but a lot of it. That contains the problem very nicely. Long haul is similar, but because drivers can only clock 10 hours a a day, a vehicle that can go for 24 has a TON of economic upside. They can cut trucking costs in half, just on labor, especially for perishable or time sensitive shipments. That’s a powerful incentive.
@Hob: It does not matter whether it is “by intent” or “by mistake” or “by happenstance”; what matters is that we’ve got a system with unstable physics. Think about a marbles and a mixing bowl. Use the mixing bowl the right way, put the marbles in the bottom, poke at them with your finger or a stick: marbles roll around, maybe go up the bowl a bit, and settle back in the bottom.
Now turn the bowl over and put the marbles on the top. Poke the marbles. What happens? Do they stay in the bowl, or do they go careening off the bowl, over the edge of the table, and roll across the floor and fall into the heating grate?
One system—the upright bowl—is stable, at least until you push past the edges of the system. The other one is unstable; it requires continual pressure to keep the marbles near the middle. Otherwise you lose your marbles.
We are, collectively, losing our marbles.
@PsiFighter37: …there are a lot of gullible people on the left…
Yes, but it’s not just gullibility; it’s a much more generalized and dangerous ignorance. We comment often about the ignorance of right wing voters, but, sadly, ignorance is rampant on the left, too. The difference is that right wingers tend to be much worse people than lefties (racists, homophobes, Trumpists, etc.). But voter ignorance is a big problem among American voters.*
*The November 5, 2019 election in Washington State offered at least one more egregious example. Other examples abound in what is, overall, a reliably blue state.