Thursday, May 30, 2013

Science Web Shows and Science Streaming

Watching the #modelorg discussion yesterday reawakened some old thoughts of mine.  Why aren't there more science panel shows?  I feel like the DeepSeaNews folks should do one, the Scientopia folks should do one, the SciAm folks should do one, etc.  There's a lot of cool things that can be done with the format, and I've a few ideas to share:

1. A Rant About Google Hangouts

Okay so Google Hangouts aren't going anywhere.  They're easy to use, require no third-party software, are free, can be streamed, and are instantly recorded for video-on-demand (VOD) purposes.  And somehow scientists I talk to and read online seem to think it's like the coolest thing ever.

But really, Hangout sucks.  I've never had a really enjoyable experience with this service.  First, there's the giant screen showing who's currently talking, while everyone else is relegated to tiny screens underneath.  You can't see people's reactions to what others are saying.  Second, every time someone new speaks, their face pops into the giant screen, while whoever was just speaking moves into the tiny ones.  This just hurts my brain.  A good panel show should aspire to be like a conversation that you're watching between people sitting around a table.  When someone new talks, their position at the table doesn't change - you just orient your vision to the new person.  Hangout destroys this spatial component and in the process disorients the viewer.  And when someone coughs or types?  Crosstalk?  You can't tell what is going on as the visual changes dominate your attention and forces your brain to lose track of the auditory input.

A better format is to give everyone equal cam space on the screen so that the viewer's eye can naturally move from one speaker to the next.  Here's a good example (pardon the StarCraft...):

Each of the four speakers gets their own separate screen real estate, and there's a fifth larger area for presenting things like figures, web pages, a camera feed from a microscope/telescope, etc.  And this is not THAT hard to do.  You will need one person on the show to own the professional version of Skype ($) so that you can do a multi-way video call.  You will need a streaming program like XSplit ($) or OBS (free) and finally an account (free) on a streaming platform (ustream, livestream,, youtube can stream as well but it only works with Adobe Flash Media Encoder Live or something I think).  It's certainly not outside the technical expertise of any scientist, but would take a few hours to get sorted out the first time.  Which means it will never happen, but hey, just throwing it out there.

2. What Makes a Good Panel Show

(a) Chemistry between the panelists.  People that know each other and are already friends work well for this purpose.  Funny people are good to have.  Interesting personalities are good to have.  A cast of four boring awkward scientists that don't know each other is not going to make a good show.  Building a repeat audience requires panelist chemistry.

(b) A small number of panelists.  Four or five are the magic numbers here.  Three is too few, and six gets either stilted or hectic, depending on the chemistry of the bunch.  If you want to incorporate more voices, have a call-in line (easily done with skype), or take text questions from chat or twitter.  By the way, ten people, the maximum in Hangout, is way way too many.  You can't get a good conversation going with ten people if everyone wants to speak (too much crosstalk) or if everyone is taking turns (not enough back and forth).  So don't do that.

(c) There must be a moderator/host!  This person is in charge of setting up the show, organizes the topics, and directs the discussion.  One thing that occurred during the #modelorg thing yesterday was that everyone would jump in with their own point, and many just ended up falling into a vacuum as the next person went on to make their own point and sometimes struggled to relate it to what other people were saying.  A moderator fixes that by asking the questions and soliciting reactions and comments on people's answers.

(d) The topic should be interesting.  Although it honestly doesn't have to be interesting on its face - if you already have a show that fulfills (a),(b), and (c), chances are you can make anything interesting.  In fact, I wouldn't really worry about the topic too much - not every online show has to be a roaring success.  Better to just learn by doing what works and what doesn't.

3. The Wider World of Streaming

If people do get into streaming more, there's a lot of content opportunities out there.  Here's a few obvious ones:
  • Streaming experiments from your lab to several classrooms watching online and answering questions from the students as you go.
  • Streaming your experiments live as you do them (where possible).  Think cameras hooked up to your microscopes/telescopes, behavioral experiments, even data analysis.  Ask the viewers to submit experiments for you to do one day.  Fun stuff like that.
  • Interview shows.
  • Debate shows.
  • Panel shows.
  • Online journal clubs.
Basically, this sort of thing is the natural next step for online communication among scientists and between scientists/science bloggers and their online audience.  So go do it!

Wednesday, May 22, 2013

Scienceifying Journalism

Here's a simple, stupid idea:

  • All journalists must now explicitly and thoroughly incorporate citations into every piece they write
  • This means every point they make where evidence, a number, a percentage, a quote, an "economists say X", an "according to current research" must be followed by a citation
  • For all interviews the date, time, duration, location, medium and interviewer should be cited
  • Where an interview is on the record, a recording and/or transcript will be made publicly available, linked from the reference section (regardless of whether the interview resulted in a quote within the piece or was just background)
  • For all citations, paragraph and line information (or timestamp information for audio/visual data) should be given so that the reader can find the exact place in the source document that the author is referring to in the piece
  • All citations and reference section will be visible by toggle only - readers can choose to read the article in a "normal" state or click a button to see the citations included

Seems common sense to me - and not that hard to implement given that we're supposed to be in the digital age now.