I don't think that adding "reactionary" would have much information-benefit; it seems like reactionaries would just be splintering off from the "conservative" category, which was already very small (just 3% of respondents last year). The vast majority (97%) of people who gave a response chose liberal, libertarian, or socialist, which suggests that the way to add information-value would be to clarify or refine those categories. (Communists were under 1%.)
There was an attempt at a more detailed political survey after last year's LW census. Its breakdown (by percent) into 12 categories came out:
31 Left libertarian, moderate non-US liberal, or "liberaltarian."
28 US liberal, progressive, or social democrat.
11 Nothing like any of those.
9 Anarcho-capitalist or minarchist, but not paleo-libertarian.
5 Libertarian socialist, anarcho-socialist, or anarcho-communist.
4 Centrist or moderate.
4 I don't care about politics.
3 Paleoconservative, paleo-libertarian, alternative right, or nationalist.
1 Fusionist conservative.
1 Green, deep ecologist, or anarcho-primitivist.
1 Marxist-Leninist.
1 Neoconservative.
1 Religious conservative.
Again, most people chose something related to liberal, socialist, or libertarian. I'm not thrilled with the particular category labels from that survey, but the results do suggest that the highest value-added would come from splitting up libertarianism into subcategories, probably left vs. right (and maybe also splitting off a category for anarchic/minarchic).
But I don't know if that would be worth the trouble to do right now. For improving the annual survey, I think it would be better to take the approach of that politics survey and run smaller-scale surveys during the year to try out different ways of asking questions, rather than flooding Yvain with hard-to-implement suggestions right before census time.
I agree that splitting up libertarianism into subcategories would likely yield some benefit. As I understand the "left vs. right" aspect of this question, the difference would mostly come down to what the person thinks about the state's role in providing social insurance. Presumably all libertarians would support a high degree of economic and social liberty -- basically letting people make decisions for themselves so long as those decisions are voluntary and they don't hurt non-consenting parties. But where "left libertarians" would be ...
The first draft of the 2012 Less Wrong Census/Survey is complete (see 2011 here). I will link it below if you promise not to try to take the survey because it's not done yet and this is just an example!
2012 Less Wrong Census/Survey Draft
I want three things from you.
First, please critique this draft. Tell me if any questions are unclear, misleading, offensive, confusing, or stupid. Tell me if the survey is so unbearably long that you would never possibly take it. Tell me if anything needs to be rephrased.
Second, I am willing to include any question you want in the Super Extra Bonus Questions section, as long as it is not offensive, super-long-and-involved, or really dumb. Please post any questions you want there. Please be specific - not "Ask something about abortion" but give the exact question you want me to ask as well as all answer choices.
Try not to add more than five or so questions per person, unless you're sure yours are really interesting. Please also don't add any questions that aren't very easily sort-able by a computer program like SPSS unless you can commit to sorting the answers yourself.
Third, please suggest a decent, quick, and at least somewhat accurate Internet IQ test I can stick in a new section, Unreasonably Long Bonus Questions.
I will probably post the survey to Main and officially open it for responses sometime early next week.