ChristianKl comments on Rationality: From AI to Zombies - Less Wrong

80 Post author: RobbBB 13 March 2015 03:11PM

You are viewing a comment permalink. View the original post to see all comments and the full post content.

Comments (94)

You are viewing a single comment's thread. Show more comments above.

Comment author: Persol 01 April 2015 01:17:22AM *  28 points [-]

Perhaps this is already discussed elsewhere and I'm failing at search. I'd be amazed if the below wasn't already pointed out.

On rereading this material it strikes me that this text is effectively inaccessible to large portions of the population. When I binged on these posts several years ago, I was just focused on the content for myself. This time, I had the thought to purchase for some others who would benefit from this material. I realized relatively quickly that the purchase of this book would likely fail to accomplish anything for these people, and may make a future attempt more difficult.

I think many of my specific concerns apply to a large percentage of the population.

  • The preface and introductions appear aimed at return readers. The preface is largely a description of 'oops', which means little to a new reader and is likely to trigger a negative halo effect in people who don't yet know what that means. - "I don't know what he's talking about, and he seems to make lots of writing mistakes."
  • There isn't a 'hook'. Talking about balls in urns in the intro seems too abstract for people. The rest of the sequences have more accessible examples, which most people would never reach.
  • Much of the original rhetoric is still in place. Admittedly that's part of what I liked about the original posts, but I think it limits the audience. As a specific example, a family member is starting high school, likes science, and I think would benefit from this material. However her immediate family is very religious, to the point of 'disowning' a sister when they found out about an abortion ~25 years ago. The existing material uses religion as an example of 'this is bad' frequently enough that my family member would likely be physically isolated from the material and socially isolated from myself. 87% of America (86% global) have some level of belief in religion. The current examples are likely to trigger defensive mechanisms, before they're education about them. (Side-note: 'Waking Up: A Guide to Spirituality Without Religion – by Sam Harris' is a good book, but has this same exact issue.)
  • Terminology is not sufficiently explained for people seeing this material with fresh eyes. As an example, ~15% of the way through 'New Improved Lottery' talks about probability distributions. There was no previous mention of this. Words with specific meanings, that are now often used, are unexplained. 'Quantitative' is used and means something to us, but not to most people. The Kindle provided dictionary and Wikipedia definitions are not very useful. This applies to the chapter titles as well, such as 'Bayesian Judo'.
  • The level of hyperlinks, while useful for us, is not optimal for someone reading a subject for the first time. A new reader would have to switch topics in many cases to understand the reference.
  • References to LessWrong and Overcoming Bias and only make sense to us.

Eliezer and Robb have done a lot to get the material into book state... but it's preaching to the choir.

Specifically what I think would make this more accessible:

  • A more immediate hook along the lines of 'Practicing rationality will help you make more winning decisions and be less wrong.' (IE: keep reading because this=good and doable) Eliezer was prolific enough that I think good paragraphs likely already exist; but need connectors.
  • Where negative examples are likely to dissuade large numbers of people, find better examples. In general avoid mentions of specific politics or religion in general. It's better to boil the frog.
  • Move or remove all early references to Bayes. 'Beliefs that are rational are call Bayesian' means nothing to most people. Later references might as well be technobabble.
  • Make sure other terminology is actually explained/understandable before it's used in the middle of an otherwise straightforward chapter. I'd try 1n & 2n-gramming the contents against Google Ngrams to identify terminology we need to make sure is actually explained/understood before casual use.
  • Get this closer to a 7th grade reading level. This sets a low bar at potential readers who can understand 'blockbuster' books in English. (This might be accomplished purely with the terminology concern/change above)
  • Change all hyperlinks to footnotes.
  • Discuss LessWrong, Overcoming Bias, Eliezer, Hanson in the preface as 'these cool places/people where much of this comes from' but limit the references within the content.

Is there any ongoing attempt or desire to do a group edit of this into an 'Accessible Rationality'?

Comment author: ChristianKl 01 April 2015 07:46:44PM 1 point [-]

I don't think the point of the sequences or the book is to be accessible to everyone. If you want to write 'Accessible Rationality' it likely makes more sense to start from stretch.

Comment author: Persol 01 April 2015 09:23:21PM 5 points [-]

Agreed that it may not be the point, but other than what I think are fixable issues, the book contents work well. I don't think starting from scratch would be a large enough improvement to justify the extra time and increased chance of failure.

I think the big work is in making the examples accessible, and Eliezer already did this for the -other- negative trigger.

"If you want to make a point about science, or rationality, then my advice is to not choose a domain from contemporary politics if you can possibly avoid it. If your point is inherently about politics, then talk about Louis XVI during the French Revolution. Politics is an important domain to which we should individually apply our rationality— but it’s a terrible domain in which to learn. Why would anyone pick such a distracting example to illustrate nonmonotonic reasoning?"