Get on your soapbox

What is public now?

For the vast majority of human history, the main way to address many people in order to share public information was to stand on a hill or a platform, in a space where as many people as possible could hear you and you could speak to them. There is obvious evidence of this scattered around the world: Red Square, Tiananmen Square, St Peter’s Square, Roman amphitheatres... Part of why this is the case is that for most of human history, whilst we’ve always had language, most humans haven’t been literate. So you had to be heard directly by them if you wanted to share a message, and didn’t want to rely on second hand relaying.

The maximum number of people you can address directly is in the tens of thousands unamplified, which, before the 20th century, was how all addresses happened. In theory, it’s since become possible to fill a giant plain or field with tens of millions of people and, with amplification, address them, but there is a limit to what you could really call ‘direct’. At some number, most people will be seeing the speaker on a screen and hearing through speakers and then it’s functionally the same as watching an address live at the cinema with others.

Television addresses then extended this to be able to address pretty much anyone who is awake to watch, which is dependent on timezone and people having televisions and electricity. The internet then extended this further, to anyone who has an internet connection and can find the address: think YouTube live streams where hundreds of millions can watch big events. Finally, if you think about asynchronous content like blog posts, viral social posts etc, then the whole connected world could see these, and it then comes down to whether they’re interested or not.

The problem with addressing more and more people with the same content is relevance. The more people you address, the less relevant it is to the average person addressed. There are exceptions to this, like for example a presidential address could be considered relevant to the whole country, but vast swathes of the population would still have no interest in hearing from their President.

Marketing has tried to address this problem of relevancy through segmentation. If you have a large potential audience like the internet, who have a near infinite amount of content and streams to ignore, then you have to be as relevant as possible and in the right place at the right time to cut through the noise. This activity, and the mastery of it, is what has made Google so valuable, having been the most valuable company in the world on multiple occasions.

However, even fine segmentation ends up crude. Broad segmentation gives significant sample size, as well as clarity of preference between one segment and another, but with lower average relevance. Fine segmentation can give higher relevance, but less clarity and certainty about preferences and relevance due to disparate segment sizes, with some very small segments. Marketers have had to balance this for a long time and optimise marketing spend in this way. Finding the right segment, channel, time, message, call to action… It’s even been called personalisation, but it’s never been personal. It’s always been about pigeon-holing you in as appropriate a hole as possible.

Now, if you think about why we’ve gone from sharing the same message to as many people as possible to segmentation in the connected world, the reason is that the idea of addressing every single person specifically and individually hasn’t been possible. Until now.

It’s possible that 1 billion people will use something like ChatGPT every day within a few years, and, if you include Google’s new AI search functionality, that may even be next year. With chat-based AI systems like this, you can address everyone personally. People are now regularly asking them questions, correcting them, continuing conversations. You don’t need to segment them to an approximate preference group. You can just know what they are directly. You could even just ask them. You can tailor every address to every single person according to their styles, likes, interests. Every product, service or content can be assessed and addressed personally for relevance and preference to that person.

Leaders, governments, companies, organisations… around the world could become more accessible to the public. Gov.chat. You could canvass voters at a never-before-seen level. Worried that a policy isn’t popular? Ask the public, the whole public, not some biased focus group, about their preference. It’s a two-way conversation too - every member of the public who wants to ask something or say something can do so, and it can be aggregated by similarity or matched against law, policy, content to be answered. The few truly novel questions then become answerable by humans.

Or, would these leaders, governments, companies and organisations learn about your fears, worries, insecurities and weaknesses, and tailor their interactions with you to manipulate you to do what they want? Buy this, vote for this, believe this, hate those people…

As usual, the truth will be in the middle.

Every influencer, thought-leader or blogger could have their own agent: Benn.agent for post-modern data realism, Joe.agent for platform-agnostic data engineering principles, Madison.agent for analytics engineering help, Zach.agent and Ben.agent for data engineering guidance, Jess.agent for SQL and remote work envy, Randy.agent for product analytics, Ananth.agent for data engineering news… David.agent for an esoteric combination of all of these, with sprinklings of AI and semantic layers, and whatever else is distracting me right now. It’s like simulating what they would say to any question or topic in your head¹, except with higher RAG-driven accuracy.

An agent can sound like me and say the same things as me, because I have already said a lot of things. I have produced a lot of text. Hundreds of thousands of tokens. Hundreds of vectors to search for relevant topics. An imprint of me, an essence of my thoughts.

1. Everyone does this, right? I can even hear the people say the things if it’s close enough to their style.