I'm working on a series of blog posts to popularize The Handbook of the Biology of Aging.

One of the dilemmas I face is between popularizing vibes vs. popularizing models. What do I mean by that?

  • Popularizing vibes trades off exactitude for getting the right feeling across. If I say "people are living longer free of disability," you feel like you know what I mean, and you know that's an encouraging sign.
  • Popularizing models makes the opposite trade. If I explain why there's a big difference between the proportion of the population with diabetes, and the length of life spent with diabetes, I have to put you in a state of icy rational focus and compromise on vibes or length.

This is LessWrong, so I expect my audience's gut reaction to be "you should popularize models!" That's my gut reaction, anyway.

But let's think about why popularizing vibes can be good. One reason is that vibes can come from a list of concrete (if slightly ambiguously stated) facts:

  • Life expectancy at birth is stalled
  • Life expectancy at 65 is improving
  • Fewer people have heart attacks and minor disabilities
  • People live longer free of disabilities, and recover from them more often
  • White men are benefitting most from getting fewer disabilities
  • We're doing a better job using medication to manage cholesterol and blood pressure

Facts are good! When you read that list, you're taking in data. It's painting a portrait of the state of American public health.

But I'm telling you these facts not because you're particuarly interested in each data point. You could look them up yourself if you were. And I don't expect you'll remember them.

Instead, I'm curating these facts because I think they'll help you form a holistic impression about whether or not Americans are getting healthier over time. So the reason I'm feeding you these facts for the vibes. And I do expect you'll remember that.

Models demand more focus to take in, and more explanation. When I explain a model, I'm trying to give you a new conceptual distinction, not just write bits into your head. If you're standing at the checkout line, browsing LessWrong on your phone, and I try to teach you a model, you're probably going to say "ugh" and read something else.

But if I can get you to understand a new model, there's a big payoff. Because now, new I can tell you about new types of facts, and therefore introduce you to a whole new vibe. You won't be able to appreciate (or overcome) the confusion in the public health literature about whether or not we're getting healthier over time unless you can appreciate the crisp distinction between population statistics and life cycle statistics. I could phrase my presentation of facts in ways that point in one direction or the other, but I would expect many of the people I want to reach to conflate them.

The aim of my post is really to teach readers to think with more sophistication about aging - to teach them the models that would let them parse the literature and practice appropriate skepticism of both "deathist" arguments against longevity research and the industry hype. The aging literature is the wild west, and mostly, you don't civilize the wild west by telling people the good and bad things about the frontier. That's popularizing vibes. You teach them the skills to navigate and survive and thrive in it, and that means popularizing models.

Any handbook is aiming for brevity - many chapters are just 8-10 pages long. It's assuming expert readers who want to build models will read the citations and teach themselves. But I'm trying to compress to 15% of the original length and still get models across in an accessible fashion to nonexpert readers.

I think some facts/vibes are still important to include, both for fun and for holistic perspective and because concrete data is useful to have. But I think I need to take a "model inventory" of each chapter and make sure I'm getting them across.

If any readers have advice on how to balance accessibility and compression while still achieving a model-communication function in their writing, let me know.

New to LessWrong?

New Comment