Web 4.0 - A web for Network States & AI

Reflections on the next web

I. 🧠

At the end of the day, what do we have?

I'll tell you what I have for you right now, and that is...

* drum roll *

A conspiracy theory!

Unlike great conspiracy theories which eventually prove themselves to be true and those who believed in it start a "We were right! Lizard-people silenced us!" campaign, this one won't prove to be true.

But in a predictive context, I believe this conspiracy theory was ahead of its time. It's slowly turning out to be prophetic, and in a few years, we're all going to be living in a reality where a version of this conspiracy theory is not only true, but considered the normal default.

This conspiracy theory is known as "The Dead Internet Theory."

The crux of the theory is this -

The Internet - as we know it, is dead. It is now primarily made up of bots, instead of real human beings. Real humans are out there enjoying their lives in the "real world", while everyone you're reading and interacting with online are actually intelligent AI bots.

This theory, like all great conspiracy theories, then continues with greater claims - that this has been built by secret cyber-military forces, funded by “The Elite” and is a deliberate attempt by the elite & big tech corporations to control the narrative and influence global culture, while keeping everyone under the illusion that who you’re talking to on the Internet is real.

Most conspiracy theorists are victims of several cognitive biases, the conjunction fallacy being one of the biggest & most obvious ones among them.

For example, let's assume that the probability of me waking up at 8am on Sundays is exactly 45%. Is the probability of me waking up on Sundays & the sun rising in the East, more or less than 45%?

If you said "more," you've just fallen into the conjunction fallacy trap. The mathematical probability of 2 events happening is always less than the probability of each of those events happening individually.

Conspiracy theorists tend to be blissfully unaware of this. In the "Dead Internet Theory" these are all the conjunctions they make -

The internet is now dead and filled with bots + These bots are actually secret AI + These intelligent bots have been created by the military for information control + Their operating costs are funded by large corporations + These large corporations are controlled by a dozen members of the global elite + They're all coordinating for a specific purpose to control you.

Obviously, this isn't true, but the reason conspiracy theories are entertaining & widely popular is that they do contain a morsel of truth. This morsel of truth is sometimes stretched too thin and wide, but a piece of mimetic truth none the less.

What is that truth?

II. 🤖

For those of us who have grown up with the Internet, those who have lived through the Web 1.0 & Web 2.0 era, we have seen the Internet go through multiple evolutions & redefine itself over the years.

What was once being used by an extremely niche audience, is now used by over 5 billion people daily. The days of StumbleUpon, niche forums & cool websites have been replaced by social media, and social media has been largely replaced by an endless algorithmic array of 30-second-video clips and 140-character-tweets.

Somewhere along the line, a lot of people are noticing that a lot of this content seem to be fake.

Content, for the sake of content, tweets that make no sense but get 1000 likes even when they're repeatedly posted, opinions that are clearly not original or well thought out getting more traction than ever, LinkedIn professionals displaying surprising lack of self awareness as they post the cringiest essays known to mankind & still being praised by dozens of NPCs in the comments, tiktok influencers dancing horribly and promoting absurd trends like licking toilets and swallowing tide pods, and so on, such that the situation is bizarre enough that it all feels like a conspiracy against the netizen's soul.

And it feels cyclical in a disturbing way, manufactured outrage at only the current inconvenience or the current “Bad Person”, with The Current Thing getting all the attention at all time, and everyone becoming experts on it overnight, and then there’s a new thing that becomes the Current Thing, and the cycle continues like an endless groundhog day with slight modifications.

You can't say you haven't noticed it, but while most of us don't give it much thought, those fond of connecting dots in 4chan forums tend to do the opposite. They double down, and the result is a glorious conspiracy theory which makes total sense until you step back and try to think rationally about it.

Most conspiracy theorists don’t do the latter.

However, while the dead internet theory is not true (yet), I'd wager that a version of it is about to come true. No, they weren’t right all along, but they have predicted what’s about to happen (yet again).

We've all seen ChatGPT demonstrate surprising levels of dialogue skills, and we'll quickly see others as Google's BARD, and Bing's Sydney demonstrate the same. In the near future when GPUs & TPUs aren't hardware bottlenecks, it'll be possible to train & run these LLMs from your iPhones. Till we get there, we'll see developers craft intricate if-else prompts that leverage these language models to mimic human behavior, or more specifically, relevant human behavior at relevant spaces. Open source alternatives will accelerate this process by 10x.

While initially limited to text, as generative media improves, these bots would make their way to other forms of media as well, whether that's through 100% indistinguishable deepfakes, or hyperrealistic simulations, it'll be soon become extremely difficult to distinguish between who is real & exists as a human being in the real world versus who is programmed to feel real but only exists in the realm of bits & bytes.

And while it all seems dystopian to us now, it'll feel extremely desirable when such technology truly takes off.

Imagine getting to talk to a version of Einstein daily, as he starts teaching you the basics of physics all the way to an expert level, while keeping the daily relationship with this AI as casual as if you're talking to a close friend. Imagine watching an AI version of your favourite teacher teaching a new course again, despite that teacher vowing never to teach on YouTube again because of the trolling comments they received.

Imagine deliberately training your own model in private, so that your grandkids' son can someday get advice directly from an AI version of you. Imagine getting access to the best doctors, lawyers, teachers, not through videos on YouTube or calls on Zoom, but as personal assistants ready to help you at any time of the day, for anything.

All these are things that I intuitively understand to be big enough casual use-cases, which means the market would incentivize this acceleration. Unless the government steps in with regulations to delay & destroy this progress, a future where humans & machine are indistinguishable online is inevitable.

Thus, the Dead Internet Theory was ahead of its time. It'll follow a chaotic organic approach instead of being top-down-elite-planned like the conspiracy theory predicts, but there would eventually be top-down winners in spaces where it's lucrative to run these intelligent bots.

Culture, may very well be programmed, specially when there are incentives to do so, and such bots will be weaponized for influence by those that profit from mass influence. This can include anyone from your friendly neighbourhood bad actor, to large corporations, to even nation states.

All of this would combine to usher in a new era of the Web, one that I'm ambitiously labelling as Web 4.0.

III. 🌍

Web 4.0 is a web which is a synthesis of the eras in the past. It is the web of Network States, of startup societies & niche token-gated aligned communities, and it is the age where everything else is AI.

The biggest theme that separates Web 4.0 from every other era is community & authenticity. In an era where the prevailing wisdom of the web is, "Not everyone you see on the Internet is real," Web 4.0 will be primarily about the separation between real people & AI bots, between aligned communities & noisy platforms, between lines of unique individuality & safe neutral generative behavior.

Here’s one mental model to think about it →

Web 1.0 -> Read Only.

Web 2.0 -> Read & Write.

Web 3.0 -> Read, Write & Own.

Web 4.0 -> Read, Write, Own & Share.

Sharing is different from trading, which is already available in Web3. You don't share your resources with someone unknown, but you do share your resources with your tribe.

This sharing doesn't necessarily mean assets. It could be something more valuable like emotional support, personal learnings & specific mentorship that people uniquely can offer to those close to them.

Consider this-

If you’re an author just starting out, you’d probably not get entry into a Network State where you have to be a bestselling author to get in. But you might get into a Network State where some authors are slightly more successful than you.

That’s more valuable as the advice they’ll give you is more relevant than what the bestselling authors can tell you. For example, bestselling authors might tell you not to focus on distribution and just keep writing, which is what works for them as they’ve built up an audience. But what slightly successful authors would tell you, and even bestselling authors would if they dialed back the clock, is that distribution matters far more when it’s early and you should be optimizing for that.

Thus, in a way, we all enter & get entry to communities that we deserve, and we’ll all upgrade our tribes and communities as we grow & decay. But the friendships we make over this journey, will likely stay with us forever.

That is the beauty of Web 4.0, an era that’s more close knit, with relationships between people are the primary feature.

Social media tried to create it, but failed when it outlawed pseudonymous accounts, monetized with advertising and then, tried to reduce churn with algorithms.

Web 3.0 NFT communities tried, but they’re failing because the primary incentive became to sell your NFT for a higher price, use newer members for exit liquidity and eventually leave your tribe behind. While pay-to-enter communities can work, it can only work when the payment signifies something crucial to the functioning of the community’s ideals and goals. The current state of price-driven communities has not enabled that.

Network States will succeed, as it’s built on a moral foundation, not a transactional one.

IV. 🗺️

Let’s look at it from another model ->

Web 1.0 -> Static Content.

Web 2.0 -> User Generated Content.

Web 3.0 -> Token-gated Holder Content.

Web 4.0 -> Token-gated Community Content.

In an age where most of the content is driven by generative content, the signal in the noise will be valuable. Just like group-chat link posting today, which is already a popular activity in various groups, Web 4.0's defining feature will be about joining communities because of the voluntary curation of links & content from like-minded friends.

In some parts of Web3, this is known as "Alpha" and reserved for token-holding members, but in true communities, the alpha is posted voluntarily and usually that’s useful to the group, and the person sharing it as they can gain insights from the conversations that occur after the conversation has been posted.

In these Web3 alpha groups, the advice might be valuable, but because there isn't any moral alignment, there’s always suspicion. Everyone knows the transactional nature of the foundation, and everyone’s aware that the person promoting the next best token is probably already an investor.

Are they looking to make you their exit liquidity? Are they trying to pull a scam before riding off into the sunset?

When a jpeg is all you can trust, and the jpeg has no moral or cultural alignment built in, it usually is a better decision to not engage. That doesn't happen in communities which aren't solely based on transactional relationships.

Hence, in network states, you'll find more value in a variety of different things than just which coins might pump 10%. It may still be token gated, but the tokens are ideally earned through contributions, rather than bought off the market.

Web 4.0 content is also geared towards fostering deeper relationships, than just parasocial relationships.

As the people in your community or network state know you & many have probably met you IRL, they understand the nuances of your content, and they trust the links & ideas you share. If you’re the subject of an online false mob attack, they might be the first ones out there defending you or supporting you, as you’d do for any of them.

That’s not the case in Web 2 or Web 3, where trust is earned through reputations built over time, and lost over trivial things like tweets.

V. 🤝

Here's another model for Web 4.0 ->

Web 1.0 -> Winners were Google, Facebook, YouTube, Twitter - ones who dominated the static & user-generated web. Web browsers can be thought of as the platform that enabled this. The various protocols like HTTPS, DNS, etc, that made it possible didn't accrue any value.

Web 2.0 -> The winners were Uber, AirBnB, Tinder, TikTok, i.e, these are the winners of the mobile web. Apple's App Store & Google's Play Store served as the platform, and they accrued value on the transactional fee layer. Meta might also emerge as the platform winner here when VR/AR apps truly take off.

Web 3.0 -> The winners are DeFi dApps, NFT communities & NFT marketplaces. Underlying protocols like Ethereum, Solana & Polkadot served as the permissionless platform & thus, also accrued value.

Web 4.0 -> The real winners would be Network States, and highly realistic AI apps. Thus, value will accrue towards platforms that enable these states, LLMs whose APIs are leveraged, or the Open Source LLM that accelerates the latter.

A lot has been written about the latter, and people have a good understanding of how AI will be the next winner of our frontier, but network states hasn’t received the same publicity. I’ll focus on that here.

Network States can begin in a lot of different ways, and they will start off being a synthesis of the above eras of the web. As every state is unique with it’s own goals, governance and ideals, there are multiple ways one can start. It may start off as token-gated morally aligned communities leveraging an on-chain ledger, or just a group chat community that’s aligned on achieving a specific goal before they organically decide to be a network state.

Communities would need a place to communicate, and Discord might emerge as the winner here, and given that they're incorporating features such as token-gating directly on the app. This is despite the "crypto = bad" pushback from the vocal minority, which shows that they're on the right path.

A lot of interesting stuff can be built just by leveraging discord bots, and I wouldn't be surprised if we see a Discord AI bot trained on a communities’ messages, serving as a spokesperson for the community & representing it publicly.

There will be decentralized alternatives for communities that value decentralization & censorship resistance as the primary goal. There might be privacy-focused communities where you might not see anything when you’re inside or any history of conversations, because everything’s auto deleted daily.

We might see platforms emerge that are built and owned by the community, specific for communities, where value is distributed among the members. Thus, just as a community isn’t a group of people in one locality, Network States may not have a single platform to coordinate.

Usually, they might start off coordinating through multiple avenues, discord being one of them, but their own platform once the community builds or funds one when it sees it necessary to do so. Till that moment arrives, they can use discord for chat, newsletters for distribution, Horizon Workspaces for VR meetups, meetup.com for physical meetups, partiful for throwing occasional citizen parties, etc.

We might also get high-friction platforms where setting up the platform or getting in is a major barrier. While this sounds like a bad thing, the entire Urbit community consists of these people, which also sends a signal of alignment among existing members who got in because they were smart enough to do so & also persistent enough to believe in Urbit's vision of a decentralized web.

It's hard to tell what will be the biggest winners of Web 4.0 network states, but it's clear that the value will accrue within the community. This isn't just limited to monetary value, but also extends to social support, dating opportunities, career options, jobs, etc. Just like the protocols of Web 1.0 didn’t accrue any value, but the protocols of Web 3.0 did, the communities of Web 1, 2, & 3 failed to accrue any ideological value (on average), but Network States of Web 4.0 will.

VI. 🚀

When you don't know who's real, and who's not, there are only 2 ways of verifying that.

The first is a hardware device that verifies Proof-of-Humanity, which can involve anything from a DNA test, a fingerprint or eyeball scanner, or something more simple like a digital activity identifier capable of proving that you’re not a bot, akin to the tech version of a nation state KYC with varying levels of effectiveness & privacy concerns.

The second is an on-chain vouching system. This is only possible in a community, where you have to vouch for the entry for a particular member. If that member misbehaves, the responsibility falls on you to be careful next time. If the next person you refer also misbehaves, your network state can decide to warn you, strip you of your referral privileges or remove you from the state, depending on the severity of the damage caused.

Such a system instantly roots out the possibility of an intelligent AI bot being let inside, as once that's discovered, the one who vouched for that bot is directly responsible. Network States can also have different human filters such as IRL meet-ups, pseudonymous video calls, etc, all of which organically happen as a community bonds together, so it's highly unlikely a bot will be able to get in and masquerade as a human for very long.

The exact opposite of this happens currently in Web3, where the leaders bring in bots themselves. NFT "communities" which are yet to mint are an example of this. You can observe this happening right now for yourself. Search for the hashtags #NFT, #Mint, #MintSoon, and find a few among the endless array of NFT projects looking to mint soon. Among those, you'll find a lot of projects having over 30,000 to 100,000 followers on twitter. Look into their discord, and you'll see an endless stream of messages on the general chat, almost everyone hyped about the NFT project, almost everyone claiming that it's the best project in the world.

All of these are bots, not intelligent bots, as is apparent from the nature of those 5 word messages designed just to signal discord activity. Those unfortunate enough to be new to NFTs might mistake such a scenario for a highly popular NFT project, thank their stars, invest in the project, only to watch their investment go to zero.

In Web3 blockchains like Solana, the cost of a transaction is so cheap and transactions are so fast that 90% of the daily active wallets are bots, actively trading across dexes trying to find arbitrage opportunities.

Who’s a bot and who’s not can be determined by tracking their wallet activity, as a normal human wallet is probably not doing 100 transactions in 5 minutes. I think a similar level of detection is possible in Web 4.0 network states, as they’re all on-chain.

So, for a user who has their wallet linked to verify identity, or interacting using it on an on-chain platform, it’ll be easy to check whether that user is displaying bot activity (mindless posts, 100 posts every few minutes) vs being a normal person.

If all else fails, IRL meet-ups or video calls can help one detect them anyway, which is also something that network states organically organize.

VII. 🏆

My idea of Web 4.0 is based on the presumption that you won’t be able to detect who is a bot and who’s not at first, or ever, but you’d eventually be able to do so if they happen to get entry into Network States or if you really start digging in and waiting for the bot to slip up. In an abundance of bots amplifying bots, the latter will be harder to do.

But this method can only work till there’s an AGI (and AGI is something completely different & superior to the LLMs we have now), at which point, every bot will be indistinguishable for humans in every aspect online. Most people have this belief that ASI is what we mean by “The Singularity”, when we’ve successfully succeeded in inventing an AI that’s superior to us.

I think, and I may be wrong on this one, that AGI is still a decade or more away. But once we get AGI, ASI won’t take 10 years or 20, as most people are predicting. To me, the timeline from AGI to ASI is more like weeks, not decades. It can even be shorter - hours, or minutes. The singularity might not hit us so slowly that it takes a year for AGI to figure out how to make ASIs.
It’s hard to speculate on something so momentous, and alignment will probably help sway the doomsday scenarios to something utopian. Regardless, it’s a black swan event, unpredictable & inevitable by design, but we have no idea how long it’ll take to get there.

Till we get AGI, I believe it is Web 4.0, as defined in this essay, that’ll define the next era of the LLM-occupied web.

If you know more about your internet friend, than you do about your next-door neighbour, you’re slowly realizing how important these online relationships are. As our barrier to the online world gets thinner & the interface gets more immersive through AR & VR, I believe we’ll all be living in communities where we share our time with like-minded individuals from all around the world. Thanks to the framework network states, we’ll eventually be able to take these relationships from the cloud to the land, and build the perfect micro-utopias to live in.

Relationships matter, and studies have shown that good relationships help us live longer & be happier, more than any other factor.

Today, we’re all at a stage where it’s possible to use technology and/or use our influence & promote the idea of network states, and do our part in building it. It could be something as grand as spending months building an on-chain token-gated social network with good UI/UX or something as little as a tweet that helps to culturemaxx this idea. (Or a retweet of this 4000-word essay.)

The only this gets built, is if enough people believe in it to acknowledge a network state as nothing more illegitimate, but perhaps superior, than a nation state. While that’ll eventually happen in a decade, we should do our part and try to accelerate it, so that we’re not not stuck wondering who is a bot and whose not.

A world where our children are free to join the states of their choosing, a world where nation state nuclear war is impossible as states aren’t confined to one geography, a world where we get old with friends who are kind, loving and share the same morals & interests as we do, is a world I’d like my children to live in.

I started this essay by asking a question -

At the end of the day, what do we have?

The answer to that is “Us”.

At the end of the day, all we have is each other.