[This is a light edit/update of a Reddit post I made about three or four years ago now.]
More than thirty years ago now, a colleague initiated a plan to terraform Mars. It is an ambitious task. Before we go there, we need vast resources. It would also be good if the 'terra' we are mimicking were a good one. As part of the overall project, I wrote up a plan for World Domination. The plan has been in the works for a couple of decades now, so if it actually takes off it would be the typical 'overnight success'.
There are numerous parts to this plan. It's too large for this space. I have been writing up separate documents for the many pieces. This is just an overview of a couple bits here. Note: all of the bits and pieces of this plan are radically affected by the rapid rise of AI. I had an OpenAI API key in 2022 before the dramatic release of the GPT that is changing the world. Even so, it caught me by surprise and by early spring 2023 I realized we were approaching a 'singularity'
This post is more about getting critiques and suggestions. Feel free to go with hyperbole. I want an idea of things I've left out, mistakes I've made, etc. However, I also want to get some idea of the trollish criticism and discouragement I'm likely to encounter along the way. I know it's going to be brutal, but I can't really anticipate the form it will take.
Quickly about me: You can find me all over the Internet without much digging. Chances are your activities actually involve open source code written by me more than twenty years ago. It is in use in hundreds of millions of devices worldwide, including my Amazon Fire TV Stick here and my Honda CRV. The code comes from a long term research program initiated in 1994. I have invented a few things under that program. Much of it is involved in this plan. I am not a genius or a superstar. I'm just a guy, but I can do stuff.
It is my intent that this scheme makes the world a better place. Personally, I am politically a hard-left socialist libertarian. I think we should all pool our resources to make the world a better place, but other than what is needed to make a life for people, I think the community should mind its own business. Most of this plan is intended to create technical mechanisms that allow people to govern themselves and know that their proxies are faithful.
Facebook is still less than ten years old. It started as a modest program written by Mark Zuckerberg. Various strategies led to explosive growth to create a social network that claims to have more than two billion users. When it the company went public, Forbes had an editorial saying that it was not worth the $75B market cap. I responded to that with an argument from the mathematics:
"Facebook is worth *more* than $75 billion and if I could purchase the whole shooting match and had the $75 billion I would put it down in a heartbeat." -- https://blog.bobtrower.com/2012/03/facebook singularity-at-1000000000000.html?q=trillion
I bring up Facebook for a few reasons. As can be seen by that article, I have some idea of why it grew so quickly and inexorably. In the article I predict a trillion dollar market cap, which at the time was absurd, but it flowed from the math.
So, I had an idea of what was happening and made a long-range prediction that was pretty solid. As someone who is familiar with data analysis, I can say that the an R2 value of .9931 is a good fit, and hence likely to be predictive. In fact, it is too good a fit and indicates to me that Facebook managed its growth with this type of thing in mind.
Facebook demonstrates something critical to belief in the feasibility of creating a large influential enterprise rapidly and certainly. Facebook at its very heart is just a small bit of software and access to commodity servers. If you could capture its user base, you could provide those other items easily enough.
Facebook faced certain challenges we do not. Facebook, and Google before them, paved the way. Google had to hack down trees and throw down gravel. Facebook had a clear path and could lay down pavement. We can simply use the existing road. It will be less resource intensive, less time consuming, and less risky.
This has been in the works for a few years, but one of my concerns was that a large competitor like Facebook or Google would swiftly crush us if we caught their attention. What has changed is that Facebook's open source code and existing API are mature and available and Facebook has bigger fish to fry with the Metaverse. At the top, they don't care if we drive down their road. They are already in the air.
For a variety of reasons, many of the top companies leave exposed flanks that would allow the capture of $1T to $2.5T market cap to a new company. One of the main reasons is discussed in this post: Trust is the New Black
Recent changes create an opening for a 'wedge' to become poised in the top areas without presenting a direct threat to exposed companies. This entre into the online universe is a news site that offers a simple proposition: It is honest and on your side. Two domains were registered for this about twenty years ago:
The first is http://VeryTrue.org, a vehicle to create an arms-length non-profit. This is intended to provide an umbrella under which to provide open source materials for code and documentation and to provide public oversight. The other domain is for its corresponding sister site, http://VeryTrue.com, the commercial entity providing content.[Right now they point to the org site]
A surprisingly simple strategy should allow the site to self-fund quickly. In addition, we have a mechanism would allow the site to capture millions of visitors without creating costs at our end.
One of the guiding principles is that we have an overall 'Zero Trust' model that makes it impossible to cheat, yet possible to audit and verify without compromising privacy. Protocols for this come from research project.
Another guiding principle is that we benefit users more than we benefit ourselves. This is still an open issue. We may leave the copyrights for things with the non-profit under a license that makes it possible for the public to police the commercial entity using the legal system.
This is a hugely ambitious undertaking and involves an enormous amount of work. Fortunately, as the plan has been forming, the infrastructure has fallen in to place funded by other companies, and open source authors have designed, built, tested, and piloted nearly all of the most difficult pieces.
The aim is to make a UI with function provided by 'plug' architecture. This would be similar conceptually to a small provably secure core kernel with drivers in user space. https://facebuxx.blogspot.com/
It is impossible to trust existing infrastructure. We will build using best practices, but that still leaves the system overall vulnerable to state level attackers. To that end, the design envisions a future move to an OS like https://sel4.systems/ running atop a RISC-V based system where all of the components are verified. These are all open source. I am not sure an existing protocol that allows verification of the design and that chips are manufactured exactly taped out. This is not yet planned, but it is anticipated that long term we should be able to verify from user to silicon that the system is completely secure.
Many things have been investigated. It is assumed that everything is part of the attack surface and that all attacks are possible. It is difficult, for instance, to secure against a 'rubber hose' attack. However, there is a strategy for this such that the user can supply credentials when appropriate, but does not know them and is physically incapable of providing them under attack. This level of extreme protection might be necessary for some individuals in charge of very sensitive credentials.
At the heart of this plan is the notion of developing a trust relationship with the majority of the online community. That means advertisements (if any) only promote things that people want (explicitly). It means that questions are answered rather than used as opportunities to exploit the person asking. It means that users have a mechanism to be entirely anonymous. It means that users can definitively withdraw their permission to use their data. It means that noxious aspects of the web like trolling, bullying, spamming, doxing etc are well contained under user control. It means that news is available appropriately, as defined by the user. It means we don't waste their time. Their time and their attention belong to them, not us.
We should be able to anticipate and answer questions that people will ask, without compromising their privacy or wasting their time. That means search that returns only the best answers you want, not what we want you to see.
Aspects such as crypto currency, voting, social networking, publishing, tools, access to copyrighted information, buying, selling, auctions, finance, services, etc. have been anticipated, but are out of scope for this already overlong post.
I am curious as to what you have to say. What have I left out? What should be done differently?
No comments:
Post a Comment