The Fifth Internet, and Why We Need It - Part 2

Part 1 can be found here https://www.dhirubhai.net/pulse/fifth-internet-why-we-need-part-1-ad-fraud-investigator/

What is the 5th Internet?

The 5th Internet is an open space filled with user atoms and content atoms. Each atom has a tail - a trust history which grows in length over time, as more trust is earned. This provides a simple and quick way of assessing the trustworthiness of a particular atom -- by observing the length of the tail. The longer the tail, the more trustworthy; the shorter the tail, the less trustworthy. Also, if there is no tail at all, then the atom was newly born; zero or nearly zero trust should be afforded it. 

Further, just like real atoms have a cloud of electrons around it, each of these atoms has a permission model and a “word cloud” around it which signals who has access to what information about the atom and what the atom is about or related to, respectively. The permission model can use the trust history of other atoms to determine what permissions are granted to other atoms seeking to interact with it. For example, if the other atom has no trust history, then no permissions are given, or the least amount of permissions are given. For example, newly created bots (user atoms with no trust history) will be given no permissions to access content, view ads, or interact with other user atoms. This is one step towards solving the huge scourge of ad fraud that is negatively impacting the digital advertising ecosystem. 

Key characteristics of the 5th Internet

Privacy. Human individuals’ privacy is the core construct of the 5th Internet and the individual can control access to and ownership of every form of their own data. They can also “signal” their demands for privacy - just like putting on clothing in the physical world - by having software agents interact with content and services on their behalf. These agents are built for the user and not built by ad tech companies - so their primary purpose and priority is to serve the human individual and not to help adtech companies make more money. 

In the 5th Internet, user atoms represent the human individual in the physical world. The user atom is the human individual’s agent in the virtual world and the permission model, word cloud, and trust history of the user atom represents key characteristics of the human individual. Because the user atom was built for the human individual, and not built for adtech companies’ profits, it helps the human individual express and enforce permissions and therefore privacy. To be specific, the privacy of an individual is violated when specific pieces of information that the user did not want to be shared, was shared with some other party -- for example the individual’s browsing history shared with adtech companies without their knowledge, consent, or recourse. On the other hand, the user willingly submits their social security number and home address when applying for a mortgage. So the key is allowing the human individual to determine what pieces of information to share with whom, and enabling them to verify that the permissions are enforced.  

Security. Security is closely related to privacy. For example, a human individual does not want hackers to break into their computer and steal their information and the human individual does not want governments listening in on messages they send. This is closely related to permissions - 1) hackers are not permitted to access the computer, 2) government snoopers are not permitted to intercept personal messages. To this end, a layer of soft-defined security called Combinatorial Security is used to protect all interactions between atoms. 

A public example of this has been available and in-use since 2013 - at https://crpt.info. It is a service that enables secure ephemeral communications. A brief, non-technical description of how Combinatorial Security works follows. When a message is sent; it is broken up into an unknown number of pieces (shards); each shard is encrypted with one of an unknown number of algorithms, and each shard is hidden in a temporary alphanumeric subdirectory of one of an unknown number of haystacks -- i.e. servers. Assuming the attacker has unlimited computing power, they would still have to find all the pieces (they don’t know how many there are, and if they found the right pieces), brute force decrypt each piece (they don’t know what algo to use, they don’t know when they are done), re-assemble all of them back together, and decrypt with the single-use private key of the intended recipient (which they’d have to steal) before the message is picked up and all the pieces evaporate. Nothing they accomplished in cracking one message helps them on the next message since everything changes, including the public-key/private-key pair (single use). User atoms and content atoms are also protected in this way. 

Trust. Finally, the most important characteristic of the 5th Internet is the trust model. In the non-logged-in expanses of the current internet, anonymity has emboldened bad actors to openly commit crimes they would otherwise not even attempt elsewhere. Further, because anyone can easily get content online, anyone can deliberately spread fake news, hate speech, and disinformation. Referring back to a point I made in Part 1, when pieces of content are abstracted from the source from which it came, it is hard to judge the trustworthiness and quality of it. For example, in the past, if the article came from a reputable source (nytimes.com, wsj.com, ft.com) a user could safely assume that the journalist did their job and an editor reviewed it before publication. But now that articles are posted to social media and all that’s left is the picture, headline, and blurb, most people don’t even look at the original source. This is what made it so easy for fake news to spread like wildfire. In 2016 a fake site abcnew.com (not the real abcnews.com) was spewing fake news that got cross posted through Facebook. Even if the user glanced at the source, they could have easily mistaken it for the real abcnews.com. 

But requiring users to log in, does not necessarily increase trust. For example, we have now observed countless cases of gamers cheating using bots -- i.e. the bots log into their account and play for the user, with perfect aim, to get 100% headshots, 100% of the time. And it would violate the privacy of users who wished to access the content without revealing who they are. So how do we solve the trust problem, while still protecting the privacy of individuals?

Right, using the trust history. You can judge the trustworthiness of a content atom or a user atom by simply looking at the length of the trust history tail. More importantly trust is earned over time, as trustworthy interactions are completed and memorialized to the “tail.” But trust can be lost instantly as well -- i.e. the entire trust history tail is chopped off when a single untrustworthy event occurs. This creates the incentive system whereby users who have spent time earning trust will think twice before doing something untrustworthy, or else lose all of the history that took time to build. Further, while most interactions only require observing the length of the trust history tail, there are some cases where an atom can give permission to another atom for a closer examination of the contents of the tail -- e.g. when applying for a passport, and real indentity information must be provided and verified. 

How does it solve the three key problems of the current internet?

Let’s start by re-stating the three problems identified in Part 1 -- 1) lack of privacy, 2) ease of fake content, including news, and 3) the dying business model of good publishers -- quality content that is ad-supported. 

It should be clear how human individual’s privacy is solved by the 5th Internet. User atoms have a permission model around them that defines who is allowed to interact with it, and what permissions are granted for each interaction. This separates the identity of the human individual from the granting of permissions, and does away with the login requirement that some systems currently use to verify trustworthiness (those didn’t work reliably anyway). To reiterate, the user atoms are not created by ad tech companies, so they act as agents that help human individuals express and enforce their wishes with respect to privacy. 

With respect to the second problem of fake content and fake news. If created in the 5th Internet, these content atoms would have zero trust history to start with. Some user atoms will interact with them but they have an easy way to tell whether to put any trust in that content atom. Over time, as other user atoms interact with this fake news content atom, their feedback will determine if the trust history grows or not; in fact, if enough users report a content atom to be fake, it can also be banned or eliminated so other user atoms will never interact with it. The exact opposite will happen for a content atom that has value. As more user atoms interact with it over time, each interaction reported as valuable and trustworthy will be memorialized to the trust history tail, building its length. This provides a simple to understand model for user atoms to judge the trustworthiness of a content atoms. 

Finally, how does the 5th Internet solve the business model problem for quality content creators? Note that monolithic companies that publish content may still go away. But individual content creators (like the journalists who worked at big publications previously) will have the ability to create content atoms and publish to the 5th Internet, and earn a living. This includes all forms of content, like music, videos, etc. The concept of micropayments comes into play here, where user atoms can make micropayments for the value they received from a quality content atom. The reason micropayments had not worked in the past is because there was too much cognitive overhead in deciding what fraction of a penny a user wanted to pay for an article they were about to read. The other problem was that a user cannot determine the value of the article before reading it (it could be a crap article, or a quality one); but once read, the user already got the value and most won’t pay afterward. In the case of the 5th Internet, micropayments are automatic, and the amount of value paid is proportional to the trustworthiness and value of the content atom, as corroborated over time by all of the other user atoms that have interacted with it previously. 

This means a quality content atom can earn value because of its quality, not because it can generate more pageviews. This eliminates the incentive to generate click-bait content that drives more pageviews, because it eliminates the entire ad-supported model that the current Internet is stuck on. Further, for YouTube creators, they can continue to create content, but the content will be incentivized to provide value to other users, instead of to drive numbers of views; this is a fundamental change because users are paying for quality content, instead of advertisers paying for number of ad impressions shown. 

So what do you think? Does the 5th Internet solve the three major, observable problems of the current internet - privacy, fake content, and business model for quality content creators. Does it solve the misaligned, and often hidden, incentives that the ad-supported Internet created? Will it eliminate the motive and opportunity to spread fake news and monetize it via ad fraud? I certainly hope so. That’s why I created the 5th Internet.