In our discussion of academic papers, Lukeprog argued that lots of smart people preferred to read ideas in academic paper format. Based on my observations, I mostly disagree. But that's just anecdotal evidence. Let's use Science!

Suppose someone at the Singularity Institute thought up a cool new idea: it could be about rationality, Friendly AI, decision theory, making money, or any of the other topics we discuss here on LW. Explaining it takes about ten pages, and it's nontechnical enough that it can be explained to a general audience of non-mathematicians. Which of the following explanations would you be most likely to actually sit down and read through?

  • A post on Less Wrong or another friendly blog
  • A book chapter, available both on Kindle and in physical book form
  • A mailing list post, made available through a public archive
  • An academic paper, downloadable over the Internet as a PDF
  • A static HTML page on the Singularity Institute's website
  • A page on a Singularity Institute or Less Wrong wiki
  • A speech, downloadable as an audio file
  • A PowerPoint or other presentation format

EDIT: To state the obvious, this poll will be biased in favor of blog postings, since it's on a blog. However, I still think it'll provide data that's much better than anecdotal guessing. I've emailed a few rationalist mailing lists to try and counteract this effect.

New to LessWrong?

New Comment
44 comments, sorted by Click to highlight new comments since: Today at 2:50 PM

Vote up this comment if you would be most likely to read a post on Less Wrong or another friendly blog.

I will say that .PDF format is end-user hostile.

[-][anonymous]12y00

RSS FTW

[This comment is no longer endorsed by its author]Reply

To state the obvious: People more likely to look at blog posts are going to be more likely to look at this blog post.

Vote this up if this poll format is terrible. An ordered list will have more information, and there are sites that let you post a poll where you can make your order preference known.

What would you propose as an alternative? LW (to my knowledge) doesn't support polls natively, and using an external site would hugely cut response rate.

I would at least make one comment saying "Vote in the options below" and then have the options as replies to it. Specify that any discussion should occur outside of that thread. The way it is now, the various poll options are scattered haphazardly throughout the comments and are hard to find.

You should also include an extra reply as a karma sink, so people can balance out the upvotes with downvotes. Not having a karma sink actually gave me a note of reluctance to vote in the poll, which suggests including one will remove one barrier to participation.

In fairness, I've never used this myself, but this

"Now, just to be clear," Harry said, "if the professor does levitate you, Dad, when you know you haven't been attached to any wires, that's going to be sufficient evidence. You're not going to turn around and say that it's a magician's trick. That wouldn't be fair play. If you feel that way, you should say so now, and we can figure out a different experiment instead."

Harry's father, Professor Michael Verres-Evans, rolled his eyes. "Yes, Harry."

"And you, Mum, your theory says that the professor should be able to do this, and if that doesn't happen, you'll admit you're mistaken. Nothing about how magic doesn't work when people are sceptical of it, or anything like that."

-- HPMOR, chapter 2

An important aspect of a scientific experiment is that you figure out the design of your experiment and how you are going to interpret the resulting data before you execute it. When you are using an experiment to resolve a disagreement, everyone involved should agree to this procedure in advance (assuming trust in intellectual honesty, which I think holds here).

Polls and surveys have self selection issues. It is good to take some step to counteract the tendency of blog readers to like reading stuff on blogs, but is it sufficient? Should Luke be convinced the bias has been remedied?

Luke was arguing in part that academic papers published in journals helps to reach a small but highly valued class of people. Should a poll that doesn't track his value of reaching the participants influence his policy decision?

I think it would be a better process, if Tom first presented the design of the poll, and allow some time for the community to critique the design. Only after modifications have been made to address criticisms, and Luke and Tom agree that it is a valid test of the question they are interested in, should the poll itself (if that is still the form of the experiment) be conducted.

I would agree if I were going to spend a lot of hours on this, but I unfortunately don't have that kind of time.

Still, I'd like to see more "measure twice, cut once" here.

If you measure once, you might have to cut twice, or you might just have to throw it away.

I expect Luke has put a lot of hours into figuring out what formats to publish information in, including gathering information about the preferences of people he wants to reach. Do you expect to do better yourself in less time?

Vote up this comment if you would be most likely to read an academic paper, downloadable over the Internet as a PDF.

For me this depends heavily on what you mean by "read"... how it depends:

I am about 3x more like to browse an academic paper all the way through than a blog post (upon seeing that the title interests me).

I am about 100x more likely to read every single word of a blog post than an academic paper (upon seeing that the title interest me).

Typically you make a "sink" post with these sorts of polls.

ETA: BTW, I went for the paper. I tend to skim blogs and then skip to the comments. I think the comments make the information content on blogs much more powerful, however.

Vote up this comment if you would be most likely to read a static HTML page on the Singularity Institute's website.

Is it worth having respondents actually weight the various options? What you are getting with the voting below is a count of the media that various people ranked first overall, but what if among those who prefer blogs (or whatever), they prefer them to pdfs (or whatever) by a small margin and among those who prefer pdfs, they prefer them to blogs by a large margin?

Also, what is the barrier to doing all or most of the items on your list?

Vote up this comment if you would be most likely to read a speech, downloadable as an audio file.

What about WATCH a speech or vlog on YouTube, given by a good public speaker (Think TED talks)? That might be my vote.

The vote is quite hard to do because it doesn't take in consideration preferences/ranking. So for myself :

  1. Blog post is what I would more likely read, if pointed to. If it's on Less Wrong, since I check it regularly, I'll very likely see it, elsewhere... depends. The good thing of blog posts is comment, both writing them and reading them is useful.

  2. Mailing list, page/static HTML : I used to read some mailing list, but I ended up just killing without reading the mails I get in them, so... if pointed to the archive I could read like a HTML page, but no more.

  3. Academic paper : I could read it, but usually "big pdf" scares me, so I need to overcome this (quite irrational, I've to admit) initial reaction, so it requires the topic to really interest me.

  4. Book chapter : I don't like e-books (especially DRM-loaded ones...) but I do read quite a lot of paper books. The problem is that I hardly will buy a book for just a chapter, so it means the other chapters have to interest me too, not just that one.

  5. PowerPoint : I almost never open those, only when like to remember things I forgot about a talk I saw in a real-life conference. But I find the format very shallow as stand-alone document, it's ok as a support to make a talk, not by itself.

  6. Audio file : requires lots of bandwidth (and my home bandwidth is saddly limited, since I'm far from the DSLAM), much harder to understand for me since English is not my native language, can't as easily stop reading for a while to ponder the idea, or re-read a part I didn't fully get, ... so the worse.

If it comes to my attention, I really don't care, as long as it's discursive written text, i.e. not audio or slides.

What are people's objections to PDF? For me PDF has two advantages. A PDF can include complex mathematics, which most papers I read do. This is generally done badly or not at all in other media. PDFs are also self-contained: I can download a PDF and have the whole thing, which is not always possible with HTML. A downside is that they don't reflow to accommodate the window size, making them awkward to read on a phone, but most HTML is broken in that way as well (including e.g. LessWrong). I've little experience of other e-book formats.

What are people's objections to PDF?

Mostly moral objections, for me.

Like what? FYI, PDF hasn't been proprietary since 2008, and there are plenty of free/open source implementations of it.

The structure of the specification is gigantic, lengthy, and wrong-headed on many points. It's a file format but it tries to define details of how it should be viewed (!!!). It is bloated and inefficient. It moves things away from becoming efficient data - plain text becomes formatted or even a flat picture of words.

As an illustration of how messed up the spec is, even Adobe Photoshop commonly fails to render PDFs the same way as Adobe Acrobat.

A PDF can include complex mathematics, which most papers I read do. This is generally done badly or not at all in other media.

This can be handled with some effort with epub and Mobi. In HTML it requires either reference to generated images or some javascript (or browser specific support).

Oh yes, there are ways of working around the limitation. But they are all just workarounds that create difficulties for either the writer or the reader, e.g. MathML is not practically writable by a human being. MathJax looks like the nearest thing to an actual solution for HTML (even if under the hood it's made of workarounds around workarounds), but I've never encountered a site that uses it other than the MathJax site itself.

Using Pandoc to convert our LaTeX math to MathJax for display sounds like the best option for us. The ability to copy and paste the generated Math could be handy. Most alternatives end up giving you images.

Using Pandoc to convert our LaTeX math to MathJax for display sounds like the best option for us.

It's worked pretty well for me so far. I don't even need to host MathJax on my site - it's one of the libraries the Google CDN provides for free. (But I don't use much LaTeX more complicated than division, natural logs, etc.)

Vote up this comment if you would be most likely to read a book chapter, available both on Kindle and in physical book form.

Everything beyond a certain length, I want to read on my Kindle; much more comfortable than a computer screen, or printed form, or my phone.

Of course, there's no reason academic papers can't be provided in multiple formats; PDF, HTML, ePUB and MOBI should all be doable. I'd like to help SI with this.

Of course, there's no reason academic papers can't be provided in multiple formats; PDF, HTML, ePUB and MOBI should all be doable. I'd like to help SI with this.

Ciphergoth, talk to me! I'm the most technical of the guys Luke has producing his papers and Luke has recently had me investigating the possibility of alternate publication formats. In particular I have looked at the production of HTML formats for display on the website. Definitely possible but based on the various obstacles that would need to be overcome and the time estimate I gave him Luke has put that side project on hold.

If you have an easier or more effective way to produce such alternate formats or want to help do it we would both appreciate it and it would be a great improvement to the website - especially if Luke also gets someone to produce some elegant and website-matching .css to go with the HTML.

I believe you have mentioned elsewhere that you (or your work) has access to some high quality file conversion tools? I was also impressed with the Python scripting you have done for grabbing the sequences.

If you want to look into this further contact me either here or at cameron.taylor [at] singinst.org. I can give you access to source LaTeX documents for you to see if your multiple output ideas will work and compare notes regarding what SingInst specific difficulties there are. If it turns out that there are things we need to do at the LaTeX template and .tex source document level in order to make multiple output possible I can take care of that - at least I am almost certain that Luke will be willing to allocate me to working with you on this to facilitate you giving us multiple output capabilities.

I'm sure ciphergoth can handle this, but Pandoc is quite well regarded, I believe. There is an online version that someone could quickly experiment with to see how appropriate it is.

(That said, I'm not sure how well it handles LaTeX input.)

I'm sure ciphergoth can handle this, but Pandoc is quite well regarded, I believe. There is an online version that someone could quickly experiment with to see appropriate it is.

Pandoc is what I have looked at so far. The results are impressive but a bunch of difficulties remain.

I like 3 of the above options for different reasons

  • I like audiofiles due to a significant commute.
  • The advantage of a post is the ensuing discussion.
  • book chapter often forces the author to be more clear than usual

Why no karma sink?

A post. If it's in a book chapter, I'd have to have acquired the book. If the book is written by someone at the SI, I'd probably hear about it here. Same thing if it's in an academic paper. I don't spend that much time on the SI's website, or pay too much attention to speeches made by its members.

[-][anonymous]12y00

.

[-][anonymous]12y00

Best use of paper: comic book / graphic novel. Looks better printed, always. Online good, on my tablet good, on paper best. Printable PDF works for this purpose.

Best use of time: blog post, which I can dump into instapaper.

I read this on a mailing list, and unsurprisingly would like to read future developments on a mailing list. :)

Or at least, that's how I'm most likely to find out such a development would exist. In practice I think I would read it on either email or a webpage.

Vote up this comment if you would be most likely to read a mailing list post, made available through a public archive.

Vote up this comment if you would be most likely to read a page on a Singularity Institute or Less Wrong wiki.