Exploration, the journey beats the guided tour.
The Over Burger Syndrome of Recommendations Algorithms
Recommendations save both time and mistakes. This, for me, is practically important when choosing a restaurant; you really don't want to meet me after poor service, mediocre food, and an expensive bill have skidded to a halt. Much better to avoid this blustering version of Phil and see a slightly larger dimensioned version after a satisfying eating experience recommended by a friend.
As many of you will note, the internet is not your friend. Certainly, it's a companion and not a bad one at times, but there's altogether too much money at stake for warmth and compassion to be the driving force behind this hyper-connectivity.
Recommendations from the various corners of the internet, then, are not driven by your friends. There was a time when this was the case. Yahoo, in the old days, provided human-driven reviews and recommendations of websites, but this has been relegated to the internet version of a weird curiosity you see in a museum that you try to work out what it actually is and how it was used.
Algorithms, those pesky things, learn from your behaviour on the internet and seek to maximise your attention by providing recommendations, content or products based on what they have identified as compelling in your context.
This is important as the internet is, in effect, an "attention economy"; your time and focus are competing over in order to maximise revenue. Slightly more hard-hitting; "If you are not paying for it, you're not the customer; you're the product being sold."
Examples;
Instagram, want your eyeballs to have more inventory (your time) in which to serve up paid-for ads.
Amazon, shows you the products you are most likely to buy or, failing that, shows paid-for ads for other products. One way or the other, you're the target and hitting the bullseye is commercially important.
Google, shows you website links deemed relevant to your immediate interest, driven by a search, and at the same time, serves up paid-for ads to drive your attention specifically to some monetisable destination.
These are just the tip of the internet iceberg, and probably the most recognisable ones at that; however, there are many, many, many more of these examples.
There are a number of issues with this, putting aside for a moment any ethical ones; firstly, recommendations become largely self-referential, feeding you more of what you have shown to be interested in narrowing down your context and driving you down a reinforcement path. Recommendations overfit.
Using my earlier restaurant example, it would be like your friend always recommending McDonalds because they've never had bad service, and you like burgers. No one is arguing (well, some are!) that Mcdonalds' doesn't have a place, but we are saying there are times when you want a different experience.
This "over burger" issue is largely why I chose to write about recommendations in this outing.
I logged into Medium the other day, which recommends articles to read; 99% of mine were about making it big as a content creator.
I'm interested in writing content, but I'm more interested in writing than "a side hustle". Here Medium has lost me; its algorithms have got it wrong, probably because Medium writers have cottoned onto the fact that there is a need for "side hustle" content and have "gone large" and filled up the medium bucket with similar tasting commodity content and secondly because the algorithm has incorrectly inferred my bias. In other words, on the input and output side, things are too narrowly banded.
The second issue with recommendations is that there isn't a "whole person" view; each website and algorithm is only shown a little bit of you, so its view of relevancy is limited. Think of it like this; you meet someone, talk to them about the weather and the person listening on the basis of one conversation believes your sole hobby is UK Weather.
Expanding recommendations to take in more data from more parts of your life and therefore become more accurate is an ethical dilemma. There's a general distaste for a commercial entity understanding all about you and exploiting it to manipulate your attention.
If we can't get past this tunnel vision of information, a recommendations solution needs in order to be (long term) useful; it stands to reason they will always have limited appeal.
We could talk about your phone as this could be the answer; it will become your recommender because you trust it enough to impart everything, and entities controlling the eco-system have a good moral compass to connect this information to useful content. But that's a more long-form article.
So, recommendations they're good at showing you more of what you know but very bad at showing you new things to explore and enjoy.
What's more, an "over burger" situation has limited appeal; you want more variety because otherwise, it becomes bland and loses its appeal.
Exploration, then, is the go-to for anyone using the internet looking to expand themselves beyond the one type of internet cheeseburger. Personal recommendations from other friends, users or like-minded individuals have great value. This is one reason I like Substack over Medium, the network effect. Rather than sneezing out a list of things to read, exploration is driven by exploring what people you like to read also like to read.
The chance of random encounters increases, and the ability to be surprised, challenged or just plain entertained is also more likely than not.
Logically this is making a case for a human relationship, driven search engine, or at the very least, a much more sophisticated way to build a network of people with whom you share bookmarks to worthwhile things. A personal version of the old-school Yahoo.
Anyway, thank you for exploring, at least in part, my thoughts on this topic. It's been a pleasure to have you along for the ride, and I apologise for the delay; I intend on a weekly outpouring of words in some digestible form, and I'm one week behind on my own goal! Life, I can recommend it, but it gets in the way of written exploration.