A Note to my New Best Friends
Chris Best and Mills Baker did the Equivalent of Holding Eye Contact with a Crazy Person on a Street Corner for One Second Too Long. Also check out Substack Notes.
Author’s Note: I have a neighbor named John who is a master carpenter who always comes over to me to give some “helpful” advice whenever he sees me doing something in my yard. It is literally my favorite thing in the world in the sense that I hate it with every fiber of my being and wish he would just stop but instead I just say “yeah” and “you don’t say?”
Chris Best and Mills Baker work for Substack and are very nice people on Notes and are actually friendly to their customers so I’m repaying that kindness by acting like John. Except I think it’s important, so I’m doing it anyway. Although I’m sure John felt the same way when one of my sprinklers started spraying into the road and he knocked on my door to let me know about this as soon as it happened.
Form Follows Function
When daVinci set out to build a flying machine, he began by looking at birds. Why? For the very practical reason that birds were the only example he knew of where flight demonstrably worked in real life. A few hundred years later, plus some thoughts about feathers and underlying principles, an airplane took flight near Kitty Hawk, North Carolina for twelve seconds.
We’ve come a long way since the Wright Brothers, but you can still look at the wing of a plane and see the spirit of a bird if you tilt your head and squint. All because regardless of the medium, form follows function and things that work tend to converge into common patterns.
This post is going to walk us through a lot of the problems we have with Trust online today and then actually bring us toward some answers. We’re going to do the equivalent of going from thinking about a bird’s wings to rigging some cables onto bits of ash and spruce so we can attempt to fly. Except we’re going to do it with Trust.
The Fundamentals of Trust
I’m close to forty now, which means I remember a world before the internet. For as long as I can remember, people have been talking about how you can’t trust what you see online. I remember Wikipedia causing my high school English teachers to laugh when we asked if we could cite it. AP Style Guides for citing websites weren’t a thing until I was close to graduation. When I was in grade school the idea of putting your credit card information onto a website was considered to be insane and unimaginable. As more and more people have started to live more and more of their lives online these concerns have morphed and evolved but they have only grown louder. We don’t worry so much about Bat Boy now. Now we worry about sedition and tyranny. We have tried to find all kind of proxies for Trust from political party to personal identity but none of them seem to work very well. None of them seem to create any kind of widespread order.
My honest belief is that this is the underlying problem that is causing the increasing division in the United States and whatever you want to call the sort of nihilistic cultural milieu in which we’ve found ourselves. Even if something divisive doesn’t play out its worst parts online, it often began there. Worse, all the solutions that are proposed typically tend to fall into the camp of “What if I just personally decided what was true and what wasn’t and impose that on everybody else by force?”
So, ask yourself: What makes you trust somebody?
Most of us spend a lot of our time online now, but turn your mind’s eye away from the screen for a moment.
History for one. Nothing beats just plain knowing somebody for a long while. Your parents and family. Your best friend since childhood. You not only know to trust them but you know how to trust them, when, where and for what.
Mutual connection is another. You say, “oh, you’re Bob’s cousin? Worked with him for twelve years!” A roofer shows up and you know he’s good because someone you know vouched for him.
Reputation is another. You’ve heard the name enough to expect a certain level of quality in some specific arena.
Audit is the last and the most work intensive one. It’s when you listen to what someone has to say, and then you get up and go out to confirm that it’s true for yourself for however long that takes and where you do whatever you have to in order to double-check the facts they’ve given to you.
Institutional Trust is Predictable
Now extract this out. What is the principle that underlies all of these that relates them all together in the same way that a Boeing 747 is connected to a Seagull?
I would argue that what underlies all of these is Prediction. Having a long history with someone allows you to understand their motives and predict what they’re going to do next. You’re always trying to understand what’s going on right now or what is going to happen next and that long history with another person helps you take what they know and use it to extend your predictions. Having mutual connections does the same thing at lower fidelity. Reputation as well. And you have so little time in your life you can never Audit what everyone else is telling you but it’s your only recourse when there’s no established trust and the truth is important. We are all basically trying to share our Audits with one another in ways that are highly accurate. The point is that underlying all of these paradigms is an attempt to take the knowledge that exists in another person’s head, figure out how well you can trust it, and then bring it into your head to form an accurate model of the world to figure out what is going on around you.
Scale this beyond the personal to groups of people.
What makes an institution trustworthy? It’s that when they tell you something is true, you can believe them because they have a long history of being honest. An institution establishes a history with you the same as your personal relationships and their one job is to continuously prove that the information they are giving to you is accurate and can be trusted. They must always demonstrate their correctness. The institution must fit together a series of facts and by following that chain of facts they put together a model of the world and explain how that model would propagate into the future. If they don’t know an answer they must tell you they don’t know. If they were wrong they must be the entity that lets you know they were wrong and then show you the amended record, explain what went wrong, and how they will stop that from going wrong again.
Also, this might sound like a given but people must believe that this is all happening. They have to really believe an institution is double-checking and being very careful before presenting something as true. If you’re doing this and nobody believes it, trust doesn’t exist. If you’re not doing this but people believe it, there can still be trust for a time until you’re wrong enough.
Trust takes a long time to build. Trust is won by inches over years and decades. People are cautious. They take small bets at first. It takes a very long while before they’re willing to put a lot of trust into an institution and actually make important decisions based on its guidance.
Conversely, trust erodes rapidly. You lose trust in miles per second. If you lie, on purpose, even one time then people will have a hard time believing anything else you ever say.
Trust Before and After the Internet
This all worked previously because it was hard to get your message out to the world. Distribution was a choke point. You could have an editorial process, review, narrative control, etc. because there was a controllable gateway for that work to be centralized. I’m not going to get into the chaos that followed the invention of moveable type but these structures evolved for a reason. People needed to believe, if only in their own minds, that the information they received was true. Otherwise, the world is too crazy to navigate. But now anyone can have a blog, or a twitter account, or a substack.
In the world of the internet, you’ll see “content” from people you don’t know, whose reputations are invisible to you, and then right then and there have to make a decision on if it’s believable. Oh, maybe it’s slightly better than that but the signal doesn’t scale to the noise. The game is more malicious, though. What gets seen the most often on the internet are things that are interesting. Attention is the only reward mechanism. Nobody is doing an audit at any scale that’s visible. So you’ll see something that by natural selection has specifically made its way to be seen by you because you want to look at it. The cold, rational part of your brain doesn’t get involved in that visibility layer at all.
There are no central choke points in the process of you viewing material online. There can’t be. You go from one site or substack to the next and it’s not like some highly trained team is vetting all of this stuff for you. There are places where information is taken in, put through that filtering process, and then presented, but it stays right there on the NYT website or insert your favorite news website. It doesn’t follow you across the internet to let you know what is or isn’t deceptive.
There are lots of scandals prior to the internet, and I’m sure you know of several but the point is that there was at least the belief, even if it was an imperfect reality, that the places people consumed their news from were performing in-depth fact checks. That simply doesn’t exist anymore. And besides that point, the sheer number of players on the field means that it is often the case that we are living with competing authority over what is true.
Begin with the Ideal End State in Mind
What would we like to be true of the news content we consume?
I’d like to know that it was vetted by someone who was an expert.
I’d like to know that it is accurate and that if it isn’t, that I will be informed of it later.
I want to know if it is really important that a lot of focus and attention was put into it and that other people will also know how important it is.
I want to know if anyone challenged it in a meaningful way.
I want to know what people who are aligned with me think of it, and what most people think of it.
I don’t want to have to go looking for all this information every single time I view a piece of content. I want to look at it and see it right there in front of me.
I don’t want to have to do a deep dive to verify every single thing myself but I would like to know that someone has done so and be able to view those records if I so wish.
I don’t want important things to be overlooked because they aren’t interesting enough.
Network Your Trust ACROSS the Internet
One of the biggest things about the way the internet works right now, that’s so big and so obvious you don’t even think about it, is that when you pull up a website your browser just renders whatever the person who wrote that website intended. You just let them have their unfiltered say on your screen. If you want to see what someone else has to say about they say, you have to leave to go to another website.
Let’s break this paradigm. Let’s turn it on its goddamn ear.
You are letting people into your virtual home when you pull up a website. And you basically say, yeah, do whatever you want. Wouldn’t you like your friends to be able to lend you their insights right away? And wouldn’t you like someone with muddy boots to take them off before walking all over your floors? These people are literally trying to find ways to sneak past your perception and trick you into reading things.
We’ll start with a desktop browser extension (I think we’re going to have to prove this works before Apple or Google will let us do it with mobile browsing). When that website loads, the url is compared against a giant database. That database looks for information from other people in your Trust Network and flags the content as it loads.
Was the headline misleading? Bam! We replace it with a better headline. You can still see the original, and there’s color-coding to make you aware of the change, but you just get a short and sweet summary with no click-bait.
Was some fact wrong? Bam! It’s annotated.
Does this journalist lie all the damn time? Bam! Their name is displayed in Red and there are all kinds of citations if you click because the name was replaced with a url.
All of these elements are easily updated and editable. All that’s missing is a database.
This doesn’t have to be negative. Imagine someone is telling you something that sounds totally crazy. Except their name is in green! You know that means they’ve never knowingly lied! You mean there really was a Syrian Brown Bear that fought in WWII during the Battle of Monte Casino?
We now have a decentralized information paradigm. It needs a decentralized editor.
Incentivize Honest Behavior and Give People Something Noble to Fight For
Now you’ll say “Different people think different things are true.”
Or “Who decides who gets to be right?”
Well, me. The King of the United States of America, Ruler of the Republic, Lord of Liberty, Supreme Commander of All Freedom, Granter of Innate Human Rights, All Powerful Separater of Powers, the Trunk which Upholds the Three Branches of Government…
Just kidding.
This will be a Republic. The thing that constantly surprises me in all of these discussion is that just letting people figure things out as a group, which has been the basis of the most successful nation ever in the history of the Earth, is never even discussed.
I’ll opt into a trust network. Me and a bunch of disaffected liberals who hate CNN more than Fox. We can talk about how to initialize those networks but I bet they grow into all sorts of tangly alliances pretty quickly after they get going anyway. I’ll passively consume its content but if I want to contribute my contribution will be reviewed by randomly assigned people within the trust network. Is what I’m saying making sense? Is it supported by the facts? That kind of thing. I don’t get to choose my own in-group reviewers and they don’t get to choose me. It’s queued and assigned totally at random.
I’ll pay some fee for my contribution the first time I do this but if it’s accepted, I’ll get that money back plus some extra for doing my work. This is to incentivize me to not cheat the system if I have something I know is wrong. It’s also to reward me for doing the right thing.
If I say something really crazy and the reviewers reject it, I don’t get my money back and it will cost me more to contribute the next time.
This also works in the inverse. If I say something accurate, it costs me less to contribute the next time and the reward is greater.
But what are we all competing for anyway? Just to tell our friends things they will likely already agree with? No, we are competing for what normal people get to see.
The free version of the Index (you can call this whatever you want) displays the consensus truth across multiple competing Trust Networks. So there are lots of people putting forward what the Truth is but only one of them gets to “win.”
If the Republican Trust Network is at war with the Democratic Trust Network both of their notes and annotations get reviewed by people randomly chosen from across multiple different Trust Networks to arrive at a consensus view.
Attention is no longer the reward mechanism of this game. Now what gets promoted is based on a game where people at base have to compete to give explanations that are convincing to other people, who don’t believe the things they believe, based on the facts of the matter at hand. There is now a decentralized editor, stretching its hand across the internet, the same way as if you had an all knowing friend who was there beside you to call bullshit whenever someone tried to tell you something nuts, or to nod along and say “yeah, man, crazy but true” in the opposite case.
There are a bunch of different rules here for different scenarios, like what to do if the consensus truth changes over some period of time.
Make this Viral by Making it Involuntary but Rewarding Voluntary Contribution
If the NYT doesn’t want to create a Trust Network, okay, but you still will annotate their web content. They can’t stop you from doing something that literally loads in someone’s browser. No matter what they have to give a url to an end user in order to load the content and they can’t just change it every day and even if they did you could just recreate the chain and link it to the content from the old url. Only literally charging the words can stop your annotations from loading because the words are what they are keyed to.
You’ll still create Trust profiles for their journalists and allow other people to use their articles as references in other arguments. They can’t stop you from referencing their available content even if it’s paywalled.
However, it would probably be nice if the NYT had a giant Trust Network and set all of their journalists to competing in what is now essentially a News Market. So you’ll make a page for every new organization that gets referenced, as well as every journalist, and you’ll send out an auto-message to whatever public email or social media they have that says “Hey, you don’t have an account, but if you did you would have made $X,XXX.XX this month.”
So you’ll tell them, “Hey, I’m still going to do this if you don’t jump in on this game. But if you do, I’ll still do the same stuff, but now you can make money for the work you were doing anyway, plus adding in some extra notes.” You’ll even get extra protections for your reputation and be able to speak in your own defense instead of having someone else do it for you.
That one article that corrects a common misconception can now get paid for every time it gets linked to a bullshit tweet (we have to have notoriety rules, but we can figure that out) that needs to be debunked.
How Would Substack Make Money From This?
People would pay you a subscription fee to have this and be part of a Trust Network. Part of that money would go to the people running the Trust Network, and part of the money would go to the participants. There’s good money there especially if you set up this tool and basically all other news organizations buy into it because you sort of have a first mover advantage and the game theory doesn’t work from the consumer standpoint to just get one version of the tool.
The real way you’re going to make money is arbitrage. The bounties on paying out for some of these stories will need to sit there and cook for a while to give time for a reasonable dispute. You’re going to park that money in some kind of low risk financial product.
Say there’s a really contentious story out there. It’s been disputed back and forth dozens of times. Again, rules get complicated, but basically the bounty on it has continued to rise and rise because the bar to believably prove it has become so high. You can’t pay out to anyone because nobody can make it through he dispute period. Let’s say someone needs to prove what actually happened to Epstein. There’s a bounty feature and people genuinely want to know this. It’s one of the rare stories in the world with a bounty of something crazy, like $10 million.
You’ve got money parked from people attempting to write this definitive story and losing. You’ve got bounty money that was set aside by people who also wanted an answer. And when someone tries to win that, you’re still going to give it something like three to six months for someone to present a reasonable challenge. Why not collect interest on that money somewhere?
So you make money on the money that people pay you. And you make money on the money that people pay you. And you use all of that to grow the business and the bounties, etc. Also there’s probably a fun and not evil way to make money on ads that doesn’t make you feel like someone just puked into your soul but this is getting long and my wife is giving me the evil eye.
In Closing
We trust because of repeated truthful answers.
This is the creation of a market for repeated truthful answers, a way of rewarding people for behaving in a civilized manner, and a way to at scale transmit those truthful explanations across the web.
If you’re worried about starting this out and not having enough content to make it worth it to people there’s a way to use chatgpt to replace headlines at scale that I’ve tested out in theory with pretty good success. That could be the starting feature.
Above all, please for the love of God build this thing. I will work for you, literally for no money and no recognition or whatever.
I know this seems like a deviation from the core product you’ve built but I don’t think it is. I think you built substack so that truth could have a place to prosper.
People will never stop arguing. They’ll never completely believe what other people say if they don’t like them. Mankind will never have full access to Truth itself. Sometimes this system will still be wrong for some amount of time. But arguments can have a direction, and by God I think that arrow can be bent to point up.
Hey there, Danilo again. I was going through my email and somehow lost yours. Would love to sync back together, I'm finally ready to bring on some collaborators. Shoot me an email at manhattanprojecttwo@gmail.com, when you have a chance. Looking forward to this.