The primary content no user wants to see any every user agent would filter out is ads. Since ads are the primary way sites stay in business, they are obligated to fight against user agents or other intermediary systems.
The ultimate problem is that Facebook doesn't want to show you good, enriching content from your friends and family. They want to show you ads. The good content is just a necessary evil to make you tolerate looking at ads. Every time you upload some adorable photo of your baby for your friends to ooh and aah over, you're giving Facebook free bait that they then use to trap your friends into looking at ads.
The only thing more insane than blaming users for having self-interest are the people who pretend that Facebook et al. are somehow owed the business model they have, painting ad-blockers as some kind of dangerous society-destabilizing technology instead of the commonsense response to shitty business practices it clearly is.
Something has to pay the bills.
When? Adblock has been free and easy for years now
Now, extensions to block ads are mature on mobile.
On some sites I run, I see 60% blocking on desktop, 10% on mobile. If it hits 60% on mobile as well, then the ads would cease to be profitable.
A quick google says somewhere from 20%-40% use adblockers. 60% sounds high, but depending on the website, I can see it. For example, I bet at least 60% of visitors to gnu.org are blocking ads and/or trackers.
But the tiny Mastodon server I run for myself, with a total user count in the low triple digits, costs about fifty bucks a month, and the users who are willing to pay cover half of that. I could probably get more of them to cover it if I was more aggressive about asking, but I prefer to keep it super low-key. I could also lower those costs if I felt like putting some work into optimizing it.
It’s not my job, it’s a thing I run on the side and put a few hours of technical work into every few months. I ain’t gonna get rich from it but it gives my friends a nice place to chat on the Internet.
I reject the notion that "We all know users won’t pay enough for a subscription service to make a huge profit that makes the VC happy, makes the founder one of the ten richest people in the country etc" because...if you can figure out how to get Grandma to use a federated Mastodon-like service, then you would do just that.
— Suzy <3”
Contrast this with real life hobby groups, and service groups, where you can more easily raise money to cover costs.
There’s a low-key buzz of occasional thanks and favors in my life that I wouldn’t get if I wasn’t doing this, either. And occasionally this connection lets one of my friends help out another who needs it, financially or emotionally, when they wouldn’t even necessarily be contact with each other, much less interested in helping out, without the shared space I’ve created.
I think this is a great experience.
(Total active user count is more like a few dozen, btw.)
I've actually experienced the latter, as I looked for an alternative to Gmail. I just found it hard to justify paying for an email provider, where the only real value add to me is the absence of ads, and not being Gmail. And really, the price is mostly irrelevant. For me to be willing to pay _anything_, it would have to have a really compelling reason to move. The value of not seeing ads is just not that high for me. And I don't think you would say there isn't value in an email provider.
If that argument sounded absurd to you, it's probably because it is. The services are valuable because people ultimately do use them — a lot of them, even. They pay for them indirectly by agreeing to look at ads.
There are loads of services we do not directly pay for, like the fire department and the public library — and yet they are immensely valuable.
I don’t agree that just because people will use a free thing that means it has a lot of value. Note I didn’t say that FB has no value, just that it might not be as valuable as one might think.
Considering most of their value is their messenger platform I don’t think FB is really worth much at all beyond their social graph.
That's a typo on my part — it doesn't change the veracity of the argument.
> I don’t agree that just because people will use a free thing that means it has a lot of value. Note I didn’t say that FB has no value, just that it might not be as valuable as one might think.
Sure, but how do you measure the "true" value? If you can answer that question, you will probably become a billionaire.
> Considering most of their value is their messenger platform I don’t think FB is really worth much at all beyond their social graph.
What are you basing this on? You may only find the messenger platform to be valuable, but how do you know how others perceive the FB platform/product?
Centralized, siloed social networks are only expensive to run because they're centralized. Things were better when the norm was to start a blog instead of using someone else's walled garden.
I'm all for empowering people. But adding personally controlled user agents to Facebook is a fundamentally flawed solution. There is no path for that to succeed because the primary content users will want to filter is ads, and the primary content Facebook needs people to see is ads. Thus user agents are an existential threat to Facebook and since Facebook controls all the content, they will ensure user agents are not allowed.
The core business model does not align Facebook's incentives with user's incentives. You can't fix that at the content level.
Doesn't that assume that there is a better business model?
Do you know of a social media company (or ad company using a product to garner the data, if that helps) that uses that model with a comparable size to Facebook? If not, are there any that are making ground?
Well, it is someone else's website. What do you expect Zuckerberg has his own interests in mind.
In 2020, it is still too difficult for everyone to set up their own website, so they settle for a page on someone else's.
If exchanging content with friends and family (not swaths of the public who visit Facebook - hello advertisers) is the ultimate goal, then there are more efficient ways to to do that without using Zuckerberg's website.
The challenge is to make those easier to set up.
For example, if each group of friends and family were on the same small overlay network they set up themselves, connecting to each other peer-to-peer, it would be much more difficult for advertisers to reach them. Every group of friends and family on a different network instead of every group of friends and family all using the same third party, public website on the same network, the internet.
Naysayers will point to the difficulty setting up such networks. No one outside of salaried programmers paid to do it wants to even attempt to write "user agents" today because the "standard", a ridiculously large set of "features", most of which benefit advertisers not users, is far too complex. What happens when we simplify the "standard"? As an analogy, look at how much easier is is to set up Wireguard, software written more or less by one person, than it is to set up OpenVPN.
The problem is that no one has the incentive to host such a service for free, and users wants the content to be available 24/7. So it's not as simple as just setting up a peer-to-peer network. Users who just use a phone as their primary computer will still want to be able to publish to their millions of followers, and so it wouldn't work to have those millions of people connect directly to this person's device. Maybe you can solve that with a bit-torrent like approach, but the problem gets harder when you include the ability to send messages privately.
You have shifted the discussion from small overlay network for friends and family to large overlay network for "millions of followers".
Those methods of sharing content with "millions of followers" are already available and will no doubt continue to be available.
A small private network is a different idea, with a different purpose. People will always have the choice of using a public network however a small overlay can avoid sending traffic through third party servers, like Facebook's.
There is no requirement that a service has to be "free", or supported by ads. This is something else you injected in the discussion. I use free software to set up overlays, but I have to pay for internet service and hosting. The cost of the "service" is not the setup it is the internet access and hosting.
It's easy for people to point what they don't like about Facebook, but I don't think you are really comprehending why they dominated the social network space to begin with. It's not as easy as making a product that doesn't advertise, if it costs money to use instead.
matrix.org is approaching this
If you mean something that can make a splash in the social media space to address the "user agents on Facebook" problem, color me skeptical about the prospect of competing on useful features with the Facebook behemoth while fighting the complexity up and down the stack of making everything decentralized and trying to make it friction-free for casual users to run, with no funding from ads, and starting out at square one on network effects. Yes, Mastodon is a possible counter to this line of argument, but Twitter is so stagnant and their product is so simple that I feel like it's almost a unique case. And for most people the Facebook use case includes being able to find all their real-life contacts; by that measure even Mastodon would fail.
The business of Facebook is not the "comically low" bar of sharing text and images with friends and family. An overlay solution that avoids sending traffic to a third party server is not "competing with Facebook". However it could be used to avoid Facebook which is the point we are discussing here. The fact that only a small number of people actually use a solution does not mean it is a "failure". If the software is relatively small, compiles fast, runs on different OS and architectures, stays available for download and reliably works as intended, then to me it is "successful". The way I evaluate "success" and "failure" of a software is probably different from many commenters/readers.
this is plain and simple fucked up.
The idea that it's okay to be a selfish child with power is tantamount to allowing driving while drunk. Power is deadly, you can just as easily crush a persons life as you could their legs with a car. Don't drive drunk, don't be in power if you can't be a responsible citizen about it.
I believe, with nothing to back this up, that social media would be improved with a better educated population. One that knows and values the basic tenets of critical thinking and debate. No amount of policing will make people smarter or more courteous. People have to choose to be more civil and be more interested in views counter to their own.
>If you have power, you have responsibility: full fucking stop.
No, if you have power you get to decide the rule.
Same as drunk driving, the reason we are not allowing driving while drunk is because the people who are against drunk driving has more power than the people who are pro drunk driving. The side that has more power get to decide what the law is.
Dictators are responsible for the death of millions, even if they make laws that say otherwise.
>Dictators are responsible for the death of millions.
The Dictators very well may believe that their action is the proper responsibility (According to him).
My point is you can't simply ask them to stop by telling them they have responsibility. Both side may view responsibility differently.
Isn't the internet supposed to be every node acting as it's own server and client simultaneously anyways? Is the problem just the inability to truly decentralize discovery, registry, and identity authentication of nodes in the network? Or is the problem that most ISPs don't want people operating services out of their homes or off of their phones?
It works for me and has worked for others. The keyword here that distinguishes this idea from almost every other "peer-to-peer overlay network" project that you can read about is "small". If you limit the size of the network, you can avoid some problems. Most projects you read about aim at the ability to create a single, large network that potentially everyone can join. Open to the public. However using a different approach it is possible to create only small networks that are only open to people you know, e.g., friends, family, co-workers. There are still problems, but there are always going to be some problems. The internet you are using right now has problems. The question is does it work well enough. The small overlay network idea has worked well enough for me that I consider it one of the better ones. It is really impossible to debate these ideas on the internet. Opinions are strong and negativity is even stronger. If you want an answer you need to try things out yourself and draw your own conslusions. No peer-to-peer solution is "perfect" and if you are always looking for the solution with zero negatives and zero limitations, you will never find it. Worse is if you never actually try these solutions, you just read about them. After you try many of them and learn what you like and do not like about the design/implementation, it is easier to chose one idea that works for your use case. Everytime someone starts promoting a peer-to-peer project you can quickly evaluate it, based on what types of designs/implementations have worked for you and which ones didn't. Well, that's my opinion, anyway.
You introduce your own idea, then shoot it down.
https://en.wikipedia.org/wiki/Straw_man
Facebook does not provide software to set up small overlay networks.
Nor is the business of Facebook to provide messaging or free web pages. That is just bait. The business of Facebook is providing a way to advertise to targeted audiences within the billions of people who visit Facebook. That is what people pay for. It is a website with billions of daily visitors.
Unless you have a webite with billions of visitors then you are not competing with Facebook.
A small overlay network is not a website, it is a computer network. It does not have a large audience because the number of nodes is small. It is an interface you see when you type ifconfig or ip addr, not an email server, a blog website or an SMS provider. You could could use it for those things or you could use it for other things, anything you can do on a LAN.
It makes zero difference what you think people want. This is not a popularity contest of any sort. If you want to argue against the premise of my comment then you need to argue that small, personally-managed overlay networks and peer to peer connections are less efficient for sharing content with friends and family than using a public website, subject to "curation" and "censorship".
This is patently not true. People were using email for this prior to Facebook's existence, which worked well enough ("Share photos via email! Share videos via email! Here's a funny story from grandma! RSVP to my birthday!") and was painless to set up; in fact, all the people who have Facebooks also have emails! But very few people are doing this anymore as their main mode of communication with friends and family; this kind of activity is now happening on Facebook, where people are happy with the ease that it facilitates. Small networks works fine, but nobody's interested, hence why Facebook is worth billions and billions of dollars and is a large website.
The stated purpose of this forum is intellectual curiosity. There is nothing in bobthepanda's comments that is directed at that stated purpose. Why is he/she arguing about email usage when the quoted sentence refers to setting up websites? This looks like more of the "straw man" argument technique. https://en.wikipedia.org/wiki/Straw_man
"no user"? Nope. People buy magazines, that are 90% ads. Subscribe to newsletters. Hunt for coupons. Watch home shopping channels. Etc, etc.
There's large part of population that wants to see ads. Scammy and bad ads? No. Good and relevant ads? A LOT of people do want them. Even tech-folks, who claim that ads are worst thing for humanity. Don't you want to learn about sale for new tech gadgets? Discounts for AWS? Good rent deals?
No, not at all. Ads don't just inform about deals, they also incite you to buy. I don't want to buy just because some random company got the chance to put some psychological manipulation in front of my eyeballs. If I want or need something, I will specifically seek it out. If I don't specifically seek it out, then I probably don't really need it.
If the price (heh) of that is that I miss out on some better deals for stuff that I would otherwise seek out, I'm fully happy eating that cost.
Sometimes we don't know that we have a problem that is easily solvable. As an example, I had no idea that there was an industry to reduce aws spending. One day a marketer made it past my filters and made me aware of an industry that I benefit from today. Just one example of many where marketing is win-win.
On the other hand I agree that there is too much psychological manipulation, unfortunately it exists because it works. Maybe this is where we need to disrupt the industry through a different approach. Just brainstorming, what if we regulated advertising to prohibit emotional/manipulative messages and rely instead on advertising facts?
I'm curious why you don't want to encounter these "manipulations". Is it because of the brainpower it uses to identify them as such? Are you particularly susceptible to giving in to them? Something else? I'm genuinely curious.
Lots of people say this online, I just don't believe them. Visual psychological manipulation works well enough on most people.
No. I want less intrusions on my senses, not more. I buy few things, and ads are irrelevant for the central ones: groceries and utilities are dependent on my place of residence.
If given the choice, I’d welcome with open arms the chance to never see an ad again anywhere, including offline. I’ve never seen an ad with a deal as good as that one, and I doubt I ever will.
Suppose there is a $100 product that produces $110 worth of value for a specific person and costs $50 to produce and deliver to them. To use economics terms, there's a $10 consumer surplus here, and a $50 producer surplus. The consumer willingness to see the ad is proportional to the consumer surplus, while the producer's willingness to pay for ad placement is proportional to producer surplus.
This is a fundamental divergence in interests; because the ad networks are paid by producers, they'll serve the ads that tend to make ad buyers the most money, not the ads that best enrich the viewers' lives.
Anecdotally of course, as one of those tech-folks who's got nothing against good and relevant ads.
Be careful what you wish for. This TED talk[0] on persuasion architectures brings up some interesting moral arguments against this idea. If we generally accept that ad conversion is a metric of an ad's success, how do we handle the situation where people that are susceptible to addictive behaviors like gambling and compulsive shopping are exposed to ads that exploit their condition? Data-driven advertising systems are built particularly well to find and exploit these kinds of people. We already see this a lot in the scam call/email world. Once you fall for one, the amount of inbound traffic you receive from scammers increases dramatically.
Ads like that used to be common. By your description, it sounds like they died out on their own.
A well targeted ad doesn't need to serve you well, it just needs to convince you to buy; I have a lot of faith in myself and I'd say those are often the same thing, but experiences may differ.
Sure at some scale, marketing becomes a requirement to build your brand for many other purposes and I will also concede that in some cases, like with services that hinge on network effect, marketing is required for the product to even become viable. But most things are not social networks and we would be better served to learn of the product in context with its competition in a less biased setting.
Sure. The thing is: when people are interested in something, they look for it. When I want to buy some games, I open my PS4's store. Any advertising I get there is absolutely fine because I asked for it. I told the software to show me the stuff that's available for me to buy.
The problem is when I open a link to some web page and 80% of my phone's screen turns into advertising noise I didn't ask for and couldn't care less about. I make it a point to delete this noise.
And yes, I know: good luck getting THAT to happen in the US given how badly funded everything else in my list is. If you’re in another country that actually funds public goods maybe this is a thing you could talk to some of your fellow techies about and make a proposal, especially if your country is getting increasingly tired of Facebook?
Alternatively, ground-up local funding of federated social networks might be workable; I run a Mastodon server for myself and a small group of my friends and acquaintances, with the costs pretty much evenly split between myself and the users who have money to spare. It is not without its flaws and problems but it is a thing I can generally keep going with a small investment of my spare time every few months.
Meahwhile, Google must be split up: e.g., Alphabet can go do AI work as a private corporation but Google Search needs to be nationalized (or split into at least 3 entities - i prefer nationalization).
Amazon must be broken up and parts nationalized, but this is a more complex case for later discussion.
B) Companies like Facebook in particular can never be nationalized due to the First & Fourth Amendments, and government being barred from competing with private entities in the private sector.
C) I do agree that Google and Amazon should be tried for anti-trust violations and that Google in particular should be split for violating anti-trust laws against vertical integration across markets.
The brits won't be happy if Google is nationalized into a french public service
Not that you're wrong, but: that's the fucking point!
Advertising delenda est.
You pay for cable. Paying customers are a better audience for ads than deadbeats.
Facebook's purpose of existence is to make money.
Do you have a point? This reads like an unfinished thought.
"mental short circuit" better describes your argument that jumps from "it's possible to both show ads and charge user fees" to "they always will".
You can do that?
Of course they conveniently only bundle these features so they can pretend it’s what customers want—hell downloading and playing when the screen is off should be a part of youtube. Just another step in the long saga of companies intentionally crippling their own services to bilk customers.
How long until those ads crop back up anyway? Instagram should give us some idea on how sponsored content might look in such a system. According to some random site, the average price for a "sponsored" instagram post is $300. You think your friends are above showing you an ad when real money is on the line? Maybe they won't be making that kind of money with very few followers, but when Pizzahut asks you to post an ad in exchange for a free pizza, I think you'll see plenty of takers. Now, granted, at least the people being paid are your friends, instead of Zuck.
And for the small group of people it seems to me that it should be easily self correcting by normal social cues since there’s no network effect to offset it.
Flaw? It seems that the point would be to force FB to transact with currency rather than a bait-and-switch tactic. The site would also be more usable if they were forced to change business model.
While everyone is sceptical on whether such a service can reach critical mass to make financial sense, a brand new FB replacement may not be able to do it, However FB itself can certainly give that as an option without hurting their revenues substantially.
I was sceptical on the value prop for Youtube Premium, I am constantly surprised how many people pay for it, if google can afford to loose ad money with YT premium, I am sure FB can build a financial model around a freemium offering if they wanted to.
The issue however is that the users advertisers care about are the ones with disposable income. The users most likely to opt out of ads are the ones with disposable income. Thus the marginal cost to Facebook from such users is significantly more than $9/quarter.
>> That is how it is today. But does it have to be like that ? What is the minimum revenue per user required for service like FB to run.
> Minimum doesn't matter, the only question is if it's more profitable than the current approach.
Only if you think strictly inside the box.
The real problem here is one is a misalignment of incentives: Zuckerberg is managing Facebook to maximize the metric he's being evaluated on (profit and wealth), not the value provided to society.
I'd start with treating users as partners, and not cattle.
Which is my point. It's easy to say "just force them to optimize for value to society" and then ignore what that really entails in practice. And what giving someone the power to do that tends to cause.
Actually solving the related problems and making things actually better is probably possible but it's messy and hard and complicated.
> Actually solving the related problems and making things actually better is probably possible but it's messy and hard and complicated.
What you describe is politics, and it's inescapable.
What does this mean in practice? How do you treat 2.6 billion people as partners?
They probably could. As they could also charge you a premium and then profit two times on top of you — with your fee and then by selling your data to third parties. Why? Because who would know that was happening? Corporations have no moral compass dictating their actions. The bottom line being what's best for investors.
20 million for YT @ 5$/month is more than 1 billion in revenue per year. Given that 20M is just 1% of the user base , there is probably not a lot impact on the ad revenue either.
It is not either free or paid , it can be both . It satisfies a need for the people willing to pay for the privilege of not seeing ads and gives additional revenue even.
No radical business changes required only a ideological change to treat their users as humans in required from FB
YouTube has been getting a lot of flack for this recently.
I am already getting such service in the form of Android's newsfeed feature on pixel. Its google but its pretty good.
You could make the same argument for Google or other online web sites relying on ads as the primary source of revenue.
Not all users hate ads in principle, just in practice. In theory, you'd be making the users select ads for relevance and not being annoying. But obviously, the site wants to show ads based on how much they're paying and "not being annoying" only factors in if pushes people off the site entirely.
The problem is actually how to fund the timeline publication services. But systems like Medium etc seem to work OK.
I am now spending several hundred dollars a year on content subscriptions. Plus subscriptions for Gmail, Zoom and a few other things where I have outgrown the free service. A freemium model for the timeline publication services would probably work.
YouTube is still reliant on ads, and ad-funded content creators, and will continue serving you manipulative content wether you pay or not. So can’t really count that as a success either.
The thing is when you remove the ugly side of those business what remains can bee done better for free using p2p networks or federation over open protocols.
Then why hasn't it happened?
Seeing how the sausage gets made certainly can make lots of people lose their taste for it.
It’s also not possible to audit the code for anything unusual without taking it all the way back through all tools source, all hardware, down through the chips, through the doping in the chips, and even lower. This stack is such a hard problem that DARPA has run programs for a long time to address this. Start by reading Thompson’s ACM article titled something like “Reflections on Trustimg Trust” where he shows code audits don’t catch program behavior, then follow the past few decades where these holes have been pushed through the entire computing stack.
From a regulatory standpoint mirroring the nature of our organizational tendencies, I posit that the _policy_ models should look similar to Mickens' security threat vector model - Not-Mossad or Mossad.
Are you implying that since we can't audit every single thing, auditing anything is useless?
No, I am pointing out that your statement implying one can know if there's anything unusual in the code can be found by an audit. It cannot. And most of the activity by big companies is not in code, it's in data, and "auditing" it is currently beyond anything on the near horizon. Some of the behavior is un-auditable in the Halting Problem sense - i.e., the things you'd want to know are non-computable.
This is so plainly false its silly. Have you ever heard of a code review? What do you think security researchers do? Google Project Zero. Plenty of things are found all the time at the higher (and lower) levels of the stack, even if something unknown remains deep within.
>And most of the activity by big companies is not in code, it's in data it is currently beyond anything on the near horizon.
Audits have no problem finding out what type of data is being collected (see PCI || HIPAA compliance). That would be a great start: for people to be made explicitly aware of all the data points that are being collected on them.
You're simply wrong. Did you read the article that shows quite clearly exactly how to do this I told you about? No, you didn't, or you'd stop making this false claim. Before you repeat this, RTFA, which I'll post again since you didn't learn last time [1].
There, read it? There's decades of research into even deeper, more sophisticated ways to hide behavior. At the lowest level, against a malicious actor, there is no current way to ensure lack of bad behavior.
>What do you think security researchers do?
Yes, I've worked on security research projects for decades, winning millions of dollars for govt projects to do so. I am quite aware of the state of the art. You don't seem to be aware of basic things decades old.
Do you actually work in security research?
>Plenty of things are found all the time
You're confusing finding accidental bugs with an actor trying to hide behavior. The latter you will not find if the actor is as big as a FAANG or nation state.
If simply looking at things was sufficient, then the DoD wouldn't be afraid of Chinese made software or chips - they could simply look, right? But they know this is a fool's errand. They spend literally billions working this problem, year in and year out, for decades. It's naive that you think simple audits will root out bad behavior against malicious actors.
Even accidental bugs live in huge, opensource projects for decades, passing audit after audit, only to be exploited decades later. These are accidental. How many could an actor like NSA implant with their resources that would survive your audits?
Oh, did I mention [1]? Read it again. Read followup papers. Do some original research in this vein, and write some papers. Give talks on security about these techniques to other researchers. I've done all that. I have a pretty good idea how this works.
>what type of data is being collected
Again, you miss. I am not talking about the data being collected. I'm talking about the data in big systems that make decisions. NNs and all sorts of other AI-ish systems run huge parts of all the big companies, and these cannot yet be understood - it is literally an open research problem. Check DoD SBIR lists for the many, many places they're paying to have researchers (me, for example - I write proposals for this money) to help solve this problem. For the tip of the iceberg, read on adversarial image recognition and the arms race to detect or prevent it and how deep that rabbit hole goes.
Now audit my image classifier and tell me which adversarial image systems it is weak against. Tell me if I embedded any adversarial behavior into it. Oh yeah, you cannot do either, because it's an unsolved (and possibly unsolvable) problem.
Now do this for every learning system, such as Amazon's recommender system, for Facebook's ad placement algorithms, for Google's search results. You literally cannot.
Don't bother replying until you understand the paper - it shows that a code audit will not turn up malicious behavior if the owner is actively trying to hide stuff from you.
[1] https://www.cs.cmu.edu/~rdriley/487/papers/Thompson_1984_Ref...
Yep!
>You're confusing finding accidental bugs with an actor trying to hide behavior. The latter you will not find if the actor is as big as a FAANG or nation state.
Wrong, wrong, wrong. Here's but one of many examples: https://googleprojectzero.blogspot.com/2019/08/a-very-deep-d...
>it shows that a code audit will not turn up malicious behavior if the owner is actively trying to hide stuff from you.
Yes, code obfuscation is a thing, but it's not a perfect silver bullet like you falsely claim, and often only serves to slow down researchers.
Of course it is true though that many bugs and vulnerabilities remain hidden which we may never find, and yet that's not a valid reason not to look for them, because there are many which are found every single day.
The perfect is the enemy of the good.
Your comment history is interestingly lacking any evidence of that. Care to demonstrate it?
>>The latter you will not find if the actor is ...
> Wrong, wrong, wrong.
Tell me how you can audit an image classifier to ensure it won't claim a specially marked tank is not a rabbit, where an enemy built the classifier, and where you don't know what markings the enemy will use. You're given the neural net code and all the weights, so you can run the system to your hearts content on your own hardware. Explain how to audit, please.
Good luck. It's bizarre you claim people can find such things, when it's a huge current research problem, and it's open if such things can even be demonstrated at all.
Same thing for literally any medium or large neural network. None can be audited for proof of no bad behavior.
>code obfuscation is a thing, but it's not a perfect silver bullet like you falsely claim,
I've never said code obfuscation. Unless you understand the paper which you seem to repeatedly ignore, you'll keep making the same mistake.
The paper demonstrates how to remove the code that has the behavior while still embedding the bad behavior in the product. You cannot find it from auditing the product code. There is zero trace in the source code. The attack in the paper has been pushed through all layers of computing stack since then, and now is at the quantum level. And as these effects become more important, there can be no audit since what you want to know runs up against physics, such as the No Cloning Theorem.
That you don't realize this is possible is why you keep making the same error that looking at source code will tell you what a product does.
If your naive belief in this were that simple there would not be literally billions of dollars available for you to do what you claim you can. DoD/DARPA/Secure foundries would love to see your magic methods.
Have you read the paper? Maybe it will show you the tip of the iceberg on why audits are used to find accidents, but are much weaker against adversarial actors, to the point of not providing any value for really complex systems.
>The perfect is the enemy of the good.
I'm not saying don't audit. I'm pointing out your initial claim that this will find anything unusual. It can find common errors. It can find bad behavior inserted by unskilled people. But against groups that know about current work, you won't find anything, in the same way you cannot audit a neural network.
That you continue to think code obfuscation is the only way to embed bad behavior in a stack shows that you're unaware of a large section of security research. Read the paper.
By "unusual" I mean anything that has intended or unintended negative effects on society, such as what was seen with Cambridge Analytica, or FBs emotional manipulation studies.
>It can find common errors. It can find bad behavior inserted by unskilled people. But against groups that know about current work, you won't find anything
Yep, and without a 3rd party audit, we can't even begin to approximate the degree of hypothetical bad behavior that exists affecting billions of people due to regular developers doing what they're told by their product managers (or of their own volition), let alone a nation state APT.
Without answering those you’re simply wasting time and effort by claiming audits can find things they cannot.
I quite doubt you work in security from your inability to grasp these things. Please demonstrate you’re not lying. Your posting history shows a tendency to be a conspiracy believer, and there’s zero evidence you do anything professionally in security, unlike the history of those I know that do work in security.
In terms of software security vulnerabilities, there is so much low hanging fruit making exploitation trivial. Even if a small team within an intelligence agency knows about a zero day deep in the stack, addressing vulnerabilities higher up in the stack that are easily exploited by script kiddies necessarily reduces attack surface.
However, what we're talking about here is not so much about security vulnerabilities, as it is about design flaws (or features) which have harmful effects on society.
It's better to incrementally address issues as they are found and weighted.
These companies do provide some value by building the infrastructure and so on. But the graph itself is kept proprietary, most likely because it is not copyrightable.
Advertising in mass media is regulated. You are very much allowed to publish claims that the government would characterize as outright lies, you just can't do it to sell a product.
If you're unpredictable you're a problem. Thus, it makes sense to slowly push you to a pole so you conform to a group's preferences and are easier to predict.
A hole in my own argument is that today's networks are incentivized to do increase engagement where a neutral agent is in most ways not.
So perhaps the problem isn't just the need for agents but for a proper business model where the reward isn't eyeball time as it is today.
But you are predictable, even if you think you are unpredictable, you are just a bit more adventurous. Algorithm can capture that as well. It will be easier for algorithm that works on your behalf.
Of course this hasn't happened yet and I doubt it ever will. Maybe I'm just insane, but most of the recommendations from services I have fed data for hundreds of hours (YouTube) are actually repulsive.
I've been on this for years. Free is a lie, and the idea that everything has to be "free as in beer" is a huge reason so many things suck.
There’s a lot of history around that split, and the motivation for HTML/CSS was about separating presentation from the content in many ways. For another example, once upon a time a lot of chat services ran over XMPP, and you could chat with a Facebook friend from your Google Hangouts account. Of course, both Google and Facebook stopped supporting it pretty quickly to focus on the “experience” of their own chat software.
The thing is that there is very little money to be made selling content, and a lot to be made controlling the presentation. So everyone focuses on the latter, and that’s why we live in a software world of walled gardens that work very hard to not let you see your own data.
There is some EU legislation proposal that may make things a bit better (social network interop), but given the outsized capital and power of internet companies i’m not holding my breath.
This was never true. There was an XMPP-speaking endpoint into Facebook's proprietary chat system, but it wasn't a S2S XMPP implementation and never federated with anything. It was useful for using FBChat in Adium or Pidgin, but not for talking to GChat XMPP users.
[1] https://googletalk.blogspot.com/2006/01/xmpp-federation.html
Everyone has email, so you could imagine a social networking app that’s just a thin layer over your email, and every interaction is encoded as an email being sent under the hood. Want to share a picture with your friends? Send an email. Someone wants to comment on it? They just send an email. Etc.
The main purpose of the app would be to offer a nice, device responsive, consistent presentation. Additionally if this were an open, documented standard, an entire ecosystem of “email apps” could emerge.
(Of course as far as your actual email account goes you’d want to auto archive the emails + not get notifications for them, but that’s easily configurable)
For example, Facebook could create some kind of plugin API that allows you to interpose your filtering/ranking code between their content and their presentation.
For example, maybe they give you a list of N possible main page feed items each with its own ID. Your code then returns an ordered list of M <= N IDs of the things that should go into your feed. That would allow you to filter out the ones you don't want and have the most interesting stuff displayed first. Facebook could display the M items you've chosen along with ads interspersed.
Something like that could run in the browser or Facebook could even allow you to host your algorithm in a sandbox on their servers if that helps performance. (Which means you trust them to actually run it, but you have to trust them on some basic things if you're going to use their service at all.)
In other words, changing the acoustics of the echo chamber doesn't mean you need to be the one implementing a big chunk of the system. You just need a way to exert control over the part you want to customize.
The RSS feeds I subscribe to give me plenty of "presentation" or "branding". Logos, written descriptions [both short- and long-form], clear names of what I am subscribing to, URLs. Just the right amount for me, in fact; if I wanted to go to their website(s) for their particular buffet of blog posts, featured puff pieces on Page Five, twitter mentions, &c I can do that ... or not. I'm glad I don't have to if I don't want to, and all of these folks are more than able to drop into their RSS feed a "Please go here for our tour information with new stuff in our online shop" mention just as you are able to go straight to some website full of deep-thumping media flashing into your senses as you get to where you want to go instead of using RSS.
There seems to be a steady thread of this sentiment here on HN, yet over the years no one has quite cracked this nut. Solutions welcome!
This is exactly why FB should be forced to federate with other social networks, by law.
This.
Also. What incentive does a walled garden even have to allow something like this? Put a different way, what incentive does a walled garden have to not just block this "user agent"? Because the UA would effectively be replacing the walled garden's own "algo curated new feed" - except if the user builds their own AI bot -- the walled garden can't make money the way they currently do.
I think the idea is very interesting. I personally believe digital UA's will have a place in the future. But in this scenario I couldn't see it working.
I was in agreement with you until I read that. People don’t need to have content dictated to them like mindless drones whether it is from social media, bloggers, AI, or whatever. Many people prefer that, though, out of laziness. It’s like the laugh track on sitcoms because people were too stupid or tuned out to catch the poorly written jokes even with pausing and other unnecessarily directed focus. It’s all because you are still thinking in terms of content and broadcast. Anybody can create content. Off loading that to AI is just more of the same but worse.
Instead imagine an online social application experience that is fully decentralized without a server in the middle, like a telephone conversation. Everybody is a content provider amongst their personal contacts. Provided complete decentralization and end-to-end encryption imagine how much more immersive your online experience can be without the most obvious concerns of security and privacy with the web as it is now. You could share access to the hardware, file system, copy/paste text/files, stream media, and of course original content.
> And isn't that how the internet used to be?
The web is not the internet. When you are so laser focused on web content I can see why they are indistinguishable.
I'm active on the somewhat (not fully) decentralized social medium Fediverse (more widely known as Mastodon, but it's more than that) and I think a lack of curation is a problem: Posts by people who post a lot while I'm active are very likely to be seen, those by infrequent posters active while I'm not very likely to go unnoticed.
How would your proposed system (that seems a bit utopic and vague from that comment, to be honest) deal with that?
If the AI is entirely under the user's control, why not? It's like having a buddie that's doing for me what I'd do for myself, if I had the time and energy (and eyebleach).
In response to it just creating more echo chambers:
- it can't be worse than now - At minimum, it's an echo chamber of your own creation instead of being manipulated by FB. There's value in that, ethically. - Giving people choice at scale means it will at least improve the situation for some people.
I am pretty sure things can be worse than right now, pretending like we are in some kind of hell state at the bottom of some well where it can't possibly be worse, seems unrealistic to me.
I think Reddit sticks strictly to your subscriptions, unless you go to /r/all.
Spoilers ahead:
In Stephenson‘s world people can hire “editors” to curate what they see, and those editors effectively determine reality for people at a mass scale. This is just one of the many fascinating ideas Stephenson explores and I highly recommend reading the book.
This interview covers some of the details if you’re not willing to dive into an 800+ page novel:
https://www.pcmag.com/news/neal-stephenson-explains-his-visi...
Part of the concept was that the agents would actually roam onto servers on the internet on your behalf raising complicated questions around how to sandbox the agent code (came in useful for VPSs and AWS-style lambdas in the end).
Or comments on different Internet forums that are blatantly spammy/SEO gaming etc.
Or block authors in search results or Twitter feed or any comment that you don't like. Basically the Zapier of content filtering.
This will be available to the user as a subscription service.
Some of these thigs are not possible on mobile platforms (Android, iOS) unfortunately because the OS do not allow such access, but we hope that Android and iOS in the future open up to allow external curation systems, apart from the app platform it's self as it's in the interest of the user.
I would love to have an agent that could apply those sorts of analyses to my data sources. In my case, I wouldn't want to filter out bad news, but unnecessarily nasty spins on it. I'd find that super valuable.
As soon as Facebook realizes you're a risk, you'll get a C&D ordering you to stop accessing their servers. These typically have the force of law under the CFAA.
You won't access their servers, but just read the page that the user already downloaded? You'll still get nailed under the Copyright Act.
"User agents" in the sense used by the OP are as old as the internet itself. There's an active, serious, and quiet effort to abuse outdated legislation to ensure that they never become a problem.
Given that users are generally not interested in exerting a high level of sophisticated control over software they use, how then is the concept of a user agent AI/filter any different at a fundamental level? It probably won't be created and maintained as a public benefit in any meaningful way, and users will not be programming and tuning the AI as needed to deliver the needed accuracy. I don't think AI has yet reached a level of sophistication where content as broad a range as what's found on the internet (or even just Facebook) can be curated to engage the human intellect beyond measuring addictive engagement, without significant user intervention.
Hopefully I'm wrong, as I do wish I could engage with something like Facebook without having to deal with ads or with content curated to get my blood boiling. Sometimes I do wonder how much it is Facebook vs. human tendency under the guise of an online persona, as both are clearly involved here.
It always appears to me that the companies are making it difficult and users have no voice.
It is important to realize that Facebook is not the first, second or even tenth of its ilk. FaceBook combines a bunch of ideas from previous systems, in particular MySpace and USENET. It is more or less the third generation of Web Social Media. There is no reason to believe there can't be a fourth.
My interest in these schemes is to provide a discussion space that is end-to-end encrypted so that the cloud service collecting the comments does not have access to the plaintext. This allows for 'Enterprise' type discussion of things such as RFPs and patent applications. I am not looking to provide a consumer service (at this stage).
The system you describe could be implemented in a reasonably straightforward fashion. Everyone posts to the timeline service of their choice and choose between a collection of user agents discovering interesting content for them to read. These aggregation services could be a paid service or advertising supported. Timeline publishing services might need a different funding model of course but bit shoveling isn't very expensive these days. Perhaps it could be bundled with video conferencing capabilities, password management or any of the systems people already pay for.
As for when the Internet/Web was not so vast. One of my claims to fame is the last person to finish surfing the Web which I did in October 1992 shortly after meeting Tim Berners-Lee. It took me an entire four days of night shifts to surf every page of every site in the CERN index.
"Time to reset the balance of power on the web and reignite its true potential.
When Sir Tim Berners-Lee invented the web, it was intended for everyone. The excitement and creativity of its early days were driven from the notion that we can all participate — and the impact was world-changing.
But the web has shifted from its original promise — and it’s time to make a change.
We can still unlock the true promise of the web by decentralizing the power that’s currently centralized in the hands of a few. How? By using the power of Solid.
Solid is the technically potent, open-source platform built to decentralize the web. Inrupt is the company that’s helping to fuel Solid’s success."
What? How will that help me or my grandma?
I think the real problem is Facebook's need to be such a large company. They brought this on themselves trying to take over the world. Maybe they need a Bell-style breakup
This is far from a perfect analogy, but compare it to the problem of email spam. People first tried to fight it with client-side Bayes keyword filters. It turns out it wasn't nearly as simple as that, and to solve a problem that complicated, you basically need people working on it full time to keep pace.
Ranking and filtering a Facebook feed would have different challenges, of course. It's not all about adversaries (though there are some); it's also about modeling what you find interesting or important. But that's pretty complicated too. Your one friend shared a woodworking project and your other friend shared travel photos. Which one(s) of those are you interested in? And when someone posts political stuff, is that something you find interesting, or is it something you prefer to keep separate from Facebook? There are a lot of different types of things people post, so the scope of figuring out what's important is pretty big.
"Holochain is an open source framework for building fully distributed, peer-to-peer applications.
Holochain is BitTorrent + Git + Cryptographic Signatures + Peer Validation + Gossip.
Holochain apps are versatile, resilient, scalable, and thousands of times more efficient than blockchain (no token or mining required). The purpose of Holochain is to enable humans to interact with each other by mutual-consent to a shared set of rules, without relying on any authority to dictate or unilaterally change those rules. Peer-to-peer interaction means you own and control your data, with no intermediary (e.g., Google, Facebook, Uber) collecting, selling, or losing it.
Data ownership also enables new frontiers of user agency, letting you do more with your data (imagine truly personal A.I., whose purpose is to serve you, rather than the corporation that created it). With the user at the center, composable and customizable applications become possible."
http://developer.holochain.org and http://holo.host Apps building using the Holochain framework/pattern: http://junto.love using
Also great are http://scuttlebutt.co.nz and https://docs.datproject.org/. Fritter is Twitter built on the DAT protocol (Paul Frazee) https://twitter.com/taravancil/status/949310662760632320
Currently, software clients of Mastodon or Twitter hold that mile. Mastodon gives all content unfiltered, which could be too much at times, while Twitter does some oft-annoying opaque black magic in its timeline algorithms.
A better solution would be to have a protocol for capability that filters content with logic under your control. A universal middleware standard that is GUI-agnostic, can fit different content types.
By adopting this, open/federated social could start catching up on content filtering features to for-profit social (in a no-dark-patterns way, benefitting user experience), hopefully stealing users.
Ideally it could be used by the likes of Twitter and Facebook—of course, given the size of for-profit social, such an integration would take some unimaginably big player to motivate them to adopt (the state of their APIs is telling), but if it’s there there’s a chance.
I do see a content filtering AI as very difficult to achieve, and I don't think it will be possible for quite some time. There are so many small problems, even getting AI to recognize targeted content is difficult, given that websites can have infinitely different layouts. And what about video or audio? The most practical way to achieve a content AI would be to persuade websites to voluntarily add standardized tags so that the only problem becomes predicting and filtering. Although I could see some issues with that like people trying to game the system.
That would be great. Having an artificial intelligence as a user agent would be perfect. That'd be the ideal browser. So many science fiction worlds have the concept of an intelligent navigator who acts on behalf of its operator in the virtual world, greatly reducing its complexity.
Today's artificial intelligences cannot be trusted to act on our best interests. They belong to companies and run on their computers. Even if the software's open source, the data needed to make it useful remains proprietary.
It’s not clear to me whether you may see posts collected by other users or only ones from your own feed and it seems highly experimental.
There is a feedback problem, though, which is that your preferences are modified by what you see. So the AI problem devolves to showing you the kind of content that makes you want to see more of it, i.e. maximize engagement. I think a lot of people are addicted to controversy, "rage porn," anger-inducing content, and these agents are not going to help with this issue.
If we could train AI agents to analyze the preferences of people, I think the best use for them wouldn't be to curate your own content, but to use them to see the world from other people's perspective. If you know in what "opinion cluster" someone lies and can predict their emotional reaction to some content, you may be able to identify the articles from cluster A that people from cluster B react the least negatively to, and vice versa. And this could be leveraged to break echo chambers, I think: imagine that article X is rated +10 by cluster A and -10 by cluster B, and article Y is rated +10 by cluster A but only -2 by cluster B. It might be a good idea to promote Y over X, because unlike X, Y represents the views of cluster A in a way that cluster B can understand, whereas X is probably some inflammatory rag.
The key is that you can't simply choose content according to a user's current preferences, they also have to be shown adversarial content so that they have all the information they need about what others think. This is how they can retain their agency. Show them stuff they disagree with, but that they can respect.
I expect that a system like the one I'm describing would naturally penalize content that paint people with opposing points of view as evil or idiots, because such content is the most likely to be very highly rated by the "smart" side and profoundly hated by the "stupid" side. Again, note that I'm not saying content everyone likes should be promoted, it's more like, we should promote the +10/-2 kind of polarization (well thought out opinion pieces that focus on ideas which might be unpopular or uncomfortable) over the +10/-10 kind of polarization (people who disagree with me are evil cretins).
So you can read more of what you already agree with? That's called living in a bubble. The mind cannot grow in a bubble.
We need access to all the data so we can decide which algorithms to apply. There’s really no escaping that
then I choose what I read in my reader
It obeys what you say. No cats? No problem. No babies? No problem.
And yes, it also does filter out ads.
FBP
I doubt FB would let you do that. It's "their" content.
Look up the "RAM Copy Doctrine". Basically, it means that every time data is copied within the computer's memory, it's a potential infringement. The HTML source of the page you've downloaded undoubtedly qualifies for copyright protection. Your argument would be an implied license to modify the page to make it suitable for display, but standard ToS will typically forbid any type of modification, removing any ambiguity into whether third-party extensions that modify the page may have an implied license. Your license would typically be limited to displaying the page as transmitted, and allowing an extension to read and modify the page would be an infringement.
I'm not a lawyer but I had a neat little SaaS business that was killed by a C&D along these same lines. Multiple attorneys reviewed and this is basically the summary. When I suggested that we do a peer-to-peer data transmission thing to avoid handling or transmitting any copyrighted content, I was warned that doing so could easily be interpreted as conspiracy and that it'd be best not to go down that route. Maybe someone else's attorney would say something different, but that's what mine told me.
There is no need for holier-than-thou censorship short of legal breaches. Good to see FB take this change of direction.
I'm not against machine curation at all, mind you. I want the infrequent poster to have a higher weighed voice and such. But I want to be able to control the parameters.
We've got this idea stuck in our heads that only the website itself is allowed to curate content. Only Facebook gets to decide which Facebook posts to show us.
What if, instead, you had a personal AI that read every Facebook post and then decided what to show you. Trained on your own preferences, under your control, with whatever settings you like.
Instead of being tuned to line the pockets of Facebook, the AI is an agent of your own choosing. Maybe you want it to actually _reduce_ engagement after an hour of mindless browsing.
And not just for Facebook, but every website. Twitter, Instagram, etc. Even websites like Reddit, which are "user moderated", are still ultimately run by Reddit's algorithm and could instead be curated by _your_ agent.
I don't know. Maybe that will just make the echo chambers worse. But can it possibly make them worse than they already are? Are we really saying that an agent built by us, for us, will be worse than an agent built by Facebook for Facebook?
And isn't that how the internet used to be? Back when the scale of the internet wasn't so vast, people just ... skimmed everything themselves and decided what to engage with. So what I'm really driving at is some way to scale that up to what the internet has since become. Some way to build a tiny AI version of yourself that goes out and crawls the internet in ways that you personally can't, and return to you the things you would have wanted to engage with had it been possible for you to read all 1 trillion internet comments per minute.