Comment by rablackburn | Hacker News Reader

Comment by rablackburn | Hacker News Reader

Parent Full thread

I agree that it seems like the problem is more a lack of reputation and trust than decentralization. But how do we build trust in a decentralized network the size of the human population? How do I know if this username on my screen is publishing in good faith without some kind of transitive trust?

I might trust someone my real-world friends trust when it comes to local gossip. If I want trustworthy national news my only recourse seems to be to go through a centralized journalism institute. But which one? Well I'll probably look at what people around are looking at.

If there's a national broadcaster or dominant cable news channel in my country I'll probably look at that, but if I don't like what I see I may be biased to go search out another institution that I find more agreeable. Then you get things like Macedonian teens running sites dressed up to look like US-based newspapers posting straight-up bullshit because there's always a market telling people what they want to hear.

How can we build a decentralized system to overcome people's biases and laziness? People are going to gravitate to sources their friends and family use, and collectively to sources that fit their preexisting beliefs. It's no wonder we're in this situation where people are struggling to agree on reality.

>> Only a decentralized approach to trust and reputation will scale well enough to regulate decentralized content

But what if a decentralized network of trust collectively agree that covid is a hoax and the Earth is flat? I'm not sure if the (de)centralization of the underlying technology really has an impact on that problem.

Transitive trust / Web-of-Trust is the solution I think. You do need to find one trustworthy entry point to such a web of trust, maybe a journalist or a news source that you trust. Once you trust one node in a network that is providing trust ratings of other nodes, you start to have some footing to judge content and authors.

Within this network, you may have a pocket of users who espouse Flat earth theory, but I believe if you were to rely on such users for trust ratings, the picture of the world created would be so incoherent, it would simply be unbearable.

> Within this network, you may have a pocket of users who espouse Flat earth theory, but I believe if you were to rely on such users for trust ratings, the picture of the world created would be so incoherent, it would simply be unbearable.

I wish I had your optimism... :(

I'm not sure I understand your pessimism. Are you legitimately concerned that, unchecked, flat earthers might actually "turn" the entire population? If everyone agrees that the earth is flat it may as well be so.

If the concern is that some people will hold mistaken beliefs then I am not sure there is a technological or sociological solution to the problem. I believe there will be people holding mistaken beliefs regardless of whether the world's information channels are "decentralized" but I agree with your parent's poster that a decentralized approach is the only approach that does not come with a high probability that the majority will be coerced or gaslit into holding false beliefs at some point in time.

> I'm not sure I understand your pessimism. Are you legitimately concerned that, unchecked, flat earthers might actually "turn" the entire population?

Not at all, but the fact is that these beliefs are growing at all thanks to the web. There are always going to be some people with completely illogical fringe beliefs. Conspiracy theories were always a thing. It's just not that despite the quality of these conspiracy theories dropping (seriously...the Earth is flat? That hasn't been a topic of debate for a loooong time), their reach and number of adherents are going up. That seems like a pretty clear regression in our society. We shouldn't be complacent about regression.

> ... a decentralized approach is the only approach that does not come with a high probability that the majority will be coerced or gaslit into holding false beliefs at some point in time.

But I don't understand how decentralized tech is somehow going to be more immune to this than the current web, which is already decentralized in most ways. Anyone can host a blog on their PC and publish what they want. We thought the blogosphere was the start of this golden age. But people aren't techies, they choose to centralise with providers like Google for gmail (email is a federated, decentralized protocol!) and Facebook for social media.

If IPFS takes off, what makes it immune from people going with the same massive provider all their friends and family are on? There's no hard barrier stopping everyone from moving off Google and Facebook tomorrow with the current web, but they won't. The big players understand this, that's why Google pays Apple billions a year to make Google the default search engine. That's all it takes to ensure everyone keeps using Google. Competition is literally a click away, people just don't click.

That said, I do support the decentralization efforts. It seems like an improvement over the current technology. I just don't buy into the hype that it's going to cure any of the problems we're seeing now. Our problem is less one of censorship, more overload of bots and bad actors.

> If everyone agrees that the earth is flat it may as well be so.

Except that it's a real, measurable regression in our knowledge as a species. I get the point you're making but that doesn't give me comfort, hence the pessimism.

Personally I am not worried that flat-earthers will turn all of us, so I am not uncomfortable, but I agree that such a regression would be quite unpalatable.

Rather, I believe that the majority of people will continue to understand the earth as "round" and will be more likely to believe the (also minority) science pedants when they say "actually, it's more slightly egg-shaped" than they will the ravings of someone who does not believe in satellites.

The beautiful trick of decentralization is not that it has any ability to prevent wrong-information from being disseminated, but that it ensures that correct-information will be disseminated regardless of any co-existing propaganda campaign to the contrary. In a centralized censorable system this guarantee cannot be met.

I believe that as long as correct-information is available the majority of humans will eventually find consensus with it.

I am not convinced that fringe ideas like flat-earthism are necessarily growing or that they would be doing so because of the web. It is likely that the web is just exposing, for the first time, the true extant of the population that is willing to accept fringe theories which while I agree is a disconcertingly large percentage, still does not seem to be the majority of people.

That said, roughly half the people I talk to still seem to fully accept and believe in some sort of literal sky god and roughly a third still seem to take in earnest the existence of actual ghosts, luck, karma, or other such poorly supported meta-think as obvious truths. But I believe this number is way way down from 50 years ago, 100 years ago, 1000 years ago... I believe more people are finding consensus with actual truth due in large part to the awesome power of decentralized messaging systems we've built to date starting with language and so far culminating at the internet and the nascent social and technical overlays that are being built on-top of it.

I’m interested in this approach, it is broadly what I’ve been wondering about since 2016.

Is anyone working on systems like this right now? Any projects come to mind?

The web already provides this: people or groups can publish under semi-stable names, linked to their real identity or not, and people can hyperlink to each other's content. The link graph is the web of trust and Google famously uses it to rank content. What is it really missing?

I guess theoretically and in some ways in practice. But I’m thinking more along the lines of “search through the web only returning results bookmarked with 2 degrees of trust” or something like that. Where each of the nodes are not documents but are identities- which may often tie 1:1 to a human

Isn't that basically newspapers? Journalists post articles and links to indicate they're at least interesting if not trustworthy, and usually under their own name. Webs of trust never worked for cryptography though.

At the very least, people would need a reliable way of indicating whether the site they are linking to is one they are endorsing or criticising, and there would need to be some clever crawler that can calculate a trustworthiness score from this graph for each page you visit.

Presumably this would be implemented as a browser extension. It could also recommend similar pages with a higher score, and maybe highly scored pages that offer a different perspective on the same topic as the page you are reading.

Google let's you link to a page without transmitting PageRank. There's no notion of transmitting negative rank because of course that would be horribly abused by competitors trying to sink each others sites.

I think the web is pretty good as a decentralised systems for ideas. A lot of basic stuff isn't exploited though. For instance a simple way to set up networks of caching proxies and point browsers towards them, so sites can survive ddos attacks and be fully anonymised would be good. CDNs like cloudflare provide that today but tools for smaller scale equivalents would be great.

I am. https://member.cash/

Thanks! I’ll check it out

Check out what Matrix are doing too:

https://matrix.org/blog/2020/10/19/combating-abuse-in-matrix...

https://secushare.org is attempting something like this. Unfortunately not even a prototype exists.

Right now, on Twitter, I can have a list of people I follow. I assume this is something like saying I "trust" them in some way.

I get a feed full of stuff from them, and stuff they've re-tweeted ie said explicitly that they trust, etc.

The whole "social network" thing really is approximately the same shape as an explicitly reified "web of trust". I would not expect the results to be terribly different, other than the added effort of having additional explicit system entities to manage.

Reply on news.ycombinator.com