I read this short piece by Wasim Salman today. It's another reminder that cyberpunk is dead because we're already living a post-cyberpunk existence. The piece doesn't make that influence subtle: it's framed using Neuromancer. At this point, direct Gibson references are futurist scripture.

A short note about a large building

There's a warehouse in Lacey, Washington State. It's 2 million square feet inside (more than 185,000 square meters, but that number isn't as startling). It's one of the largest buildings in the world. Target uses it to store imported merchandise prior to distribution.

Here it is on a Google Map. Zoom out a few times and you'll still be able to see it more clearly than anything else around.

It's not your (segmentation) fault

Hi there. It's time for a quick market segmentation break. I'm talking about how marketers categorize their audience into different groups, i.e., young people vs. old people or single people vs. families. This segmentation can get very high-resolution once behavioral data enters the picture. You've likely seen those surveys about soda brands and political affiliations.*

For a quick look at your own local market segmentation, plug your ZIP code into this Nielsen site.
My neighborhood is rich in Up-And-Comers (younger people with college educations, no kids, and hybrid Nissan Altimas). If you've seen a lot of targeted advertising in recent years, it's not entirely your fault. Blame your neighbors.

There are some clear comparisons to be made here between marketing demographics and microbial ecology. The taxonomy tends to be more clear-cut when it comes to microbial species, but both microbes and humans occupy specific niches for specific reasons. I'd also guess that, much as microbial cultures usually involve more than one species,** human societies rarely match any specific market segment. If they did, the segmentation model wouldn't be terribly useful.

I tend to be inherently distrustful of marketing but its methods could offer some novel insights into microbial communities.

*Also relevant from that survey: the question "Is Olive Garden Authentic?"

**Plus phage!

Of paper mountains and mysterious broths

Nature has had a few articles lately about the most highly-cited research papers in existence. This infographic is part of the most recent analysis. The whole context can get a bit silly so I'm glad they approached it from more of a popular-science direction than a genuinely metatextual one. I'm also not a fan of most infographics so it's nice to see a clean, compact figure like this one or the interactive figure in the main article.*

There are few surprises here: the most frequently-cited papers are those offering novel scientific methods or easy implementations of those methods. The Altschul BLAST papers are a great example. They describe sequence comparison methods which are so easy to use and powerful that everyone from undergraduates to senior researchers still find them useful on a daily basis. Of course, methods eventually become common knowledge and people either stop citing them, replace them with newer methods, or just forget who created the methods in the first place. Laemmli buffer, described in a 1970 paper by its namesake,** is still used in proteomics studies, but so is LB medium, a recipe originally described as "lyosgeny broth" but often called "Luria-Bertani" medium after its creator, Giuseppe Bertani, and the microbiologist Salvador Luria.***


*Apologizes if you can't access content behind the paywall! I'm not sure if this article is broadly accessible but it really ought to be.

** Yes, that's right, the seminal paper by Dr. Buffer.

***The 1951 paper is here. This 2004 review offers a nice historical perspective.

I read this Newsweek article about negative results this morning. It's not bad as far as mainstream science and medicine reporting goes, though that's admittedly a low bar to clear.

It's obvious that negative results need to be published. The urgency is especially clear when those negative results could have direct implications for health.

Even so, I was struck by the metaphor used in a JAMA article the Newsweek piece quoted:

“In baseball, it is easy to find out just how well Cal Ripken has hit against various pitchers in the past, at home or away games, in recent weeks or during his career,” Dickersin and Rennie wrote. “Yet in medicine, there is no comprehensive source for finding out similar, accurate statistics for medical interventions. How can baseball be better organized and keep better records than medical science?”

The comparison really triggers some knee-jerk pedandry for me. Yes, the authors are being a bit tounge-in-cheek. It still isn't a fair comparison at all. Baseball has clearly defined rules which haven't changed much over the last century. We can compare batting averages from 1914 with those from 2014 and understand what the values mean in both contexts. There really isn't a way to do that with medical treatments other than whether patients lived or died (even that is a moving target, and a recursive one at that since medical science impacts life expectancy). Truly useful long-term results may take decades to obtain. Most baseball games don't take that long.

I just found this software called Beaker today - it's essentially a way to mash together several different data analysis and presentation languages, allowing output from one to seamlessly become input for the next. To be fair, this isn't too difficult to do manually as long as the data sets are properly organized, but it's frequently a pain to convert something like R output to a presentable format without a few extra steps. Beaker appears to handle that. It'll even export to LaTeX as far as I can tell, so I can put off learning that for another year or so!

I haven't had a chance to try it out yet, but if Beaker is really as helpful as it seems then it could really save me some time. It would be nice to automatically export sets of R code and output to pretty HTML, at least.

I should have gone to Fire Investigator school: The Work Importance Profiler

I took another work-related quiz this past week: the Work Importance Profiler. You can take it yourself here. This kind of survey was once called the Minnesota Importance Questionnaire, but despite what the name may imply, it's not about how folks feel about Minnesota.* The survey determines your values in a work context. The theory is this: you may sacrifice your skills and interest to work at an easy, uninteresting job, but you will rarely sacrifice your values. Those values include who you become as part of your job, what you get, who you get to know, and how others regard you.

Here are my results:

OK, to be fair, weighting values like this may be misleading. Working Conditions may rank below Achievement in this list but I still wouldn't work at a job with a genuinely terrible work environment (let's picture a BSL-4 virology lab without proper ventilation, for example). I do tend to value accomplishment above all else. I would never want a job where I don't feel like I'm accomplishing anything.

I'm not sure if it's part of the standard survey, but the values can be converted into career categories. Here are my top 10 matches:

Ah, so I should be an actor! These results don't appear to correlate with those from interest surveys like the Strong so I have to take them with a grain of salt.


*I imagine it would look like this.