The Decentralized Post-AI Web

The Web is becoming decentralized, and it has nothing to do with crypto and blockchains.

The social media platforms, work tools, and websites we visit are becoming less and less universally adopted. The internet has delivered on its promise of connecting everyone to everything, but it turns out everyone actually only wants to be connected to a few things.

As a result, the web is becoming increasingly siloed and niche.

The overall internet is trending towards a more decentralized model where people are moving away from large, general social media platforms like Facebook (a.k.a. Web 2.0).

Instead, we increasingly interact in smaller, niche communities such as specialized forums, private groups, and dedicated apps/tools.

The novelty of global interaction is wearing off, and people are once again craving small, more personalized communities.

This new, more fragmented web results in "echo chambers." Rather than connecting us, the web is pulling us further apart.

This is evident through the rise in popularity of Left-leaning social media platforms like Bluesky, Threads, etc. and Right-leaning social media platforms like Truth Social, Gab, etc. Rather than bringing us together, we're all congregating in hyper-personalized echo chambers that never challenge our views, ideas, or opinions. Personally, I spend more time interacting with micro-social networks within Discord and Slack than I do on big public social media platforms.

You could say this is a good thing because it builds closer relationships than you get with pseudo-anonymous social media, where you're one in a billion. An argument could also be made that large social media platforms are echo chambers controlled by watch time maximizing algorithms and who you follow (TPOT, anyone?).

The merit of that as it relates to society is another discussion, but what interests me more is that the same thing is happening at work, too.

With really good LLM AI systems, you no longer have to think, "I wish someone would build this tool!" and hope it shows up one day on Product Hunt. Now you just say, "Hey ChatGPT, build this." The crazy part is that, for the most part, it builds it.

Where previously there was an industry consensus around the tools most commonly used to solve a problem, now anyone from the janitor to the president of the company (or, I suppose, even the country) can build and use a micro tool in mere minutes.

There are a few reasons why this is for sure a really bad thing:

An unrivaled cyber security disaster. This is shadow IT but on crack.
What are you going to put on job listings and resumes? No one has any meaningfully demonstratable experience if everyone only uses micro tools at work.
A maintenance mess. Tech debt is generated faster than any human can type.
And many more...

But, there are reasons why this is a really good thing:

It allows for creating tools that otherwise wouldn't be worth creating.
Fewer SaaS bills.
A solution for the exact problem at hand. No more, no less.
Opportunities for self-directed efficiency increases across the org chart.
Rapid ideation and prototyping
And many more...

I'm not going to pass judgment on whether long-term it is good or bad (it is probably both), but I personally will, of course, take advantage of any opportunity I can (while trying to mitigate the negatives).

So, bringing this back, socially, we're growing further apart on the internet, and professionally, we're having less and less in common. The decentralization of work and society isn't something I would have predicted. Rather than an interconnected web, we all have our own little webs that are loosely connected with little overlap.

As it becomes easier and easier to grow apart, I think it is important that we prioritize working together and understanding everything that everyone else has to offer. We shouldn't get stuck reinventing the wheel just because we can. There are benefits to scale and community that we won't and can't get in our own little worlds.

The tech community has learned this lesson many times. Now, rather than writing everything bespoke, we use and create libraries. We've created sayings like D.R.Y. (do not repeat yourself) to discourage creating duplicate logic. We shouldn't give ourselves too much credit, though; we didn't come to this idea naturally. Instead, it was a lesson learned through failure brought on by the mess of code we wrote and couldn't maintain. While there are benefits of bespoke solutions, truly unique problems are rare. I don't want to make a prediction, so I won't, but it will be interesting to follow this trend and see if we, as a society and as professionals, continue down this path of decentralization or once again converge on common ground.