I've had poorly-considered rationalizations for why I under-hyperlink many of these posts. It was partly because I think "my target audience", to whatever extent that means anything, is already aware of e.g. Chesterton's Fence, or at least followed the link in my previous piece that mentioned it, before they saw it in the headline yesterday. And part of it is because people should be more ready to google things while they read – lead forward, hands on keyboard, this is not a drill.
But once I gave it some thought, the more I think there's an actually good reason, aside from all the rationalization. It's partially about web literacy – the idea that we should read laterally, or horizontally, pulling up other sources and seeing what's said about the topic more generally. But it's also a concern that, if you followed my links, you'd allow me to contextualize my own writing far too much.
Don't get me wrong, I can google "Bayesian update", and link in the appropriate wikipedia article for you. But they'll be little gaps – why is that page about inference and not updates specifically? What the important part? Is the math important? I could tell you my answers, based on my usage – but isn't that cheating you out of a moment to consider my writing critically? Is it nonsense? Is it so vague as to be meaningless? Does the specificity of the link not apply to the circumstance I present it in?
At the end of the day, no author should be interpreted only using their own (preferred) contextualization. It gives them too many weaselly ways out of making a point that can stand or fall on its own. It also lets them fail to communicate by not trusting their audience to read the content well. If a term is confusing, you should google it, and build your own durable, independent understanding – if an idea is confusing, that's bad writing. The flow of the text is interrupted either way, but if you just lean back and click you might not catch the worst flaws.
Of course, almost all the stuff I remember writing for this website is basically coherent. But I'm approaching post #200 – some things I've written must be misleading or in desperate need of a new context – and because I want good ideas, not just to promote my own thoughts, I need to build a readership more likely to help along those lines. In many ways, I invite confirmation bias if I supply too much of the data. See the reasoning, and when you're surprised, grab the data you think answers the question.
In some ways, this is basic media literacy stuff, but given some of the feedback I've gotten, I think it's fair to say that bad media literacy in a readership seems to be heavily incentivized. Judge that how you will, but I'd prefer not to be a part of it.