Weekly Head Voices #18: Refactor my dogfood.

Welcome all, to this, the 18th edition of the Weekly Head Voices, in which I discuss a number of issues that mostly have nothing specific to do with the 11th week of 2010, but which might or might not have crossed my mind during that time! Issues include good news on the EuroVis 2010 front, a new edition of the Head Voices Review featuring my completely unexpected stay in a 7 Tesla MRI scanner (as a test subject, of course)  and finally some nerdy backyard philosophy dealing with the well-known itch to Rewrite Everything From Scratch, Because What’s There Now Sucks.

First, because I have no other visual element for this week’s post, and  because I am, as you might have noticed, a method blogger, I present you with probably the best chatroulette.com improv I’ve ever seen so far. For those of you who have been completely asleep the past few weeks, chatroulette is a new site that’s been taking the interwebs by storm. The site pairs up random strangers for webcam chats. One is allowed to move to the next random stranger with the click of a buttom (resulting in the new English verb “to next” someone..). Random hilarity (and often perversity) ensues! Here’s that mostly SFW and brilliant piano improv:

YouTube Preview Image
(The original video had to be pulled, after 4 million views, due to some YouTube issues. Merton, pianist, has posted this new version. Thanks to Francois for the heads-up!)

Back to business, we have just heard the great news that our two EuroVis 2010 submissions have been finally accepted for publication.  The articles are:

  1. S. Busking, C.P. Botha, and F.H. Post, “Dynamic Multi-View Exploration of Shape Spaces,” Computer Graphics Forum, 2010.
  2. P.R. Krekel, E.R. Valstar, J. de Groot, F.H. Post, R.G. Nelissen, and C.P. Botha, “Visual Analysis of Multi-Joint Kinematic Data,” Computer Graphics Forum, 2010.

Fantastic work you first authors you! Remember people: YOU HAVE TO CITE THESE SOON, AND YOU HAVE TO CITE THEM OFTEN! This also means that a number of us will be going to Bordeaux in June (I can imagine worse places to go to in June) to mingle with other scientists and to drink really good red wine.

That good news brings us to the influential WHV feature, the Head Voices Review! (My TPN is still working on the new jingle.  Hopefully it’ll be on time for our joint Vodka review feature.) In this edition, I’ll be reviewing the Philips 7 Tesla MRI scanner and the JBL Duet-200 computer speakers.

On Friday, I unexpectedly had the pleasure of trying out a state-of-the-art 7 Tesla Philips MRI scanner, as a test subject. I can report that the bore, although small, is quite comfortable. However, test subjects with claustrophobic tendencies should probably look elsewhere.  Scanning can be quite noisy, especially when a diffusion weighted imaging protocol is applied that involves scanning in 162 different gradient directions (to study the structural connectivity in my brain). However, the music that gets piped in between scanning sessions more than makes up. After scanning, it was scientifically confirmed that there is indeed a brain housed in my skull, an observation that pleasantly surprised me. Soon I hope to be able to post visualisations, made by one of our MedVis ninjas, of the structural connections in my brain.

All in all, if you have a few million euros lying around, this piece of kit is highly recommended.  To summarise:

  • Philips 7T MRI: AWESOME.

In a previous review, I was quite negative about the Logitech S3-30 speakers, for a large part due to the absolute mess of cables that it comes with. This past week I took delivery of the new JBL Duet-200 speakers I ordered, for the grand amount of 30 eurobucks, to replace it. The JBL is a single unit containing two speakers, something which might be seen as a drawback, but which I consider an advantage. In spite of its compact appearance, it packs quite a base and more than sufficient volume.  In addition, there are exactly two (2) cables: One for the power supply, and one for the audio. I can only hope that Logitech contracts JBL on their next PC speaker product design. My only (minor) gripe is that the JBL-supplied audio cable is only about 40cm long. All in all:

  • JBL Duet-200 at € 30 price-point: AWESOME.

MRI scanners and PC speakers: No product too big, no product too small for the Head Voices Review!

Finally, it’s time for some backyard philosophy.  This week, it’s a brief point of really nerdy philosophy, although I think the principles apply to some non-nerdy activities as well. Software developers, as well as many other engineering types, often reach a point during working on a project, when they have a hard-to-control urge to trash the whole thing and start from scratch.  There is usually a very strong belief that the project / software / product can be designed much better by starting from scratch.

It turns out that this is an insidious and mostly incorrect belief, for a large part due to all the knowledge present in the existing “ugly” product that will get thrown out. Engineers easily underestimate the importance of this knowledge. It turns out, much as we don’t like to hear this, that refactoring is, nine times out of ten, a far better answer than rewriting from scratch. Experienced developers know this, and are mostly able to suppress the rewrite from scratch itch.

Joel Spolsky wrote a really good essay on this phenomenon. Go read it!

4 Comments

  • What about operating systems? Windows NT and MacOSX seem to have become quite successful. I guess it depends on the software and circumstances. (Yes, I am defending our choice to write a new dtitool from scratch.. although it is only the base system and for the plug-ins that implement the actual functionality we will re-use “old” code).

    • Agreed, it depends on the circumstances, which is also why I covered myself with “nine times out of ten”. :)

      However, I do think that sometimes the choice for a rewrite is too easily made. I’m sure you had very good reasons to do this with DTITool, plus that you are in fact reusing a great deal of old code.

  • That is the beauty of stats, in this case if you rewrite, there is a90% (assuming cpb’s stats are correct) that you are wrong… moral of the story… I know it is boring but refactor! (or get someone else to- and call it a life lesson!)
    Re : 7T … sooo much of Telsa… eish.

  • BTW, check out Ben Fold’s response to the video above, http://www.youtube.com/watch?v=LfamTmY5REw see the full text from http://music-mix.ew.com/2010/03/22/ben-folds-chatroulette%E2%80%94merto/

Join the Discussion

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>