Few people choose to fill their homes with bright fluorescent lighting. Perhaps in the kitchen or utility room where bright lighting is helpful for the kinds of things they do in those rooms, but not in the living room, study or bedroom, where they wish to feel comfortable, relaxed and secure.

On the pages Neuroscience and Implications for Software Engineers I describe how feeling comfortable, relaxed and secure - rather than stressed and anxious - enables us to use the faculties we need to be good at programming.

So logically, we should light programming shops in the same way that we light our studies at home. I believe that commercial organizations drift into doing exactly the wrong thing because raising stress and anxiety has an insidious and previously unsuspected allure. The neurochemical response to stress - raised dopamine and norepinephrine - is similar to the response to addictive drugs!

I’ve recently been working in a shop that’s good in several ways, and one of their good practices is low level lighting in several areas where programmers have their desks. Some of the old-timers have told me that the lighting is an accident of history, but I don’t care how it got that way. It’s still worth sharing!

It’s a bit tricky sharing a lighting level - darned modern cameras :-) So in the two photos below, notice how strong the illumination seems where the lighting shines directly on the walls. And remember I took them to make a point, not as works of art. So they’re “warts and all”, and therefore I’ve taken care to avoid catching anything identifiable in them. The areas used for greeting customers are much smarter:

Compare with these two shots of the kitchen area, where natural daylight can be seen coming through the windows:

The low level lighting brings several interacting benefits. The sense of cosiness, similar to my living room at home, is certainly there. Beyond that, there’s an improved sense of privacy. This is more of a perceptual illusion than a real effect, because we are all as visible to each other as in any open plan environment. But it feels more private, and that’s what matters for reducing stress and anxiety and so enabling juxtapositional thinking.

There’s also an effect similar to being in a library. People respect the sense of quiet and cosiness. They keep their voices down, and tend to go elsewhere to use their phones. There’s a nice kitchen they can go in for a chat. Not all noises are the same, and we know that noise containing intelligible data (even if it’s not intelligent) is the worst kind for breaking our concentration and preventing us seeing complex pictures. So if we must suffer open plan, anything which encourages people to keep their voices down is good. The office depicted in the photos above has a babble annoyance level which is nearly as good as a two person office. It’s remarkable. I would never have guessed this effect would be so pronounced if I hadn’t had the chance to observe it happening.

Then there’s eye strain. We all spend our days looking at screens, which emit their own light. They don’t need external illumination like pieces of paper do. In fact, the more environmental light there is, the harder we find it to see the screens. The blatant extreme case of that is glare when the angles are wrong, but flicker has a more serious effect on its victims, wearing them down over time. The flicker from the kind of cheapskate tubes found in many commercial shops is bad enough, but when it beats with displays which are refreshing at nearly the same frequency it can be (almost) visible to most people. There are plenty of people who think that computer displays emit noxious rays which give them headaches when actually it’s that beat frequency flickering away at the edge of their awareness that’s doing them in. Simply explain what’s happening to them and they stop being so annoyed by it - try it! Obviously, the less fluorescent light there is in an office, the less these effects will occur.

The whole flicker thing has another side to it. Many of the best programmers are the sort of people who are classified as being on the “autistic spectrum”. Now I’m very suspicious about such classifications. As I explain on the page Other Applications, I believe we can understand the name-calling as unconscious resentment by those trapped in the psychosocial stress/dopamine economy towards that don’t participate and exchange fixes. The greater sensitivity of the non-participants can be understood as normal sensitivity, not blunted by excess dopamine constantly sloshing around their brains. But whatever the cause, many of the best programmers can see the 50 or 60 Hz flicker of fluorescent lighting without any monitors beating away nearly in sync, and it’s extremely annoying to them. Reducing it improves their ability to concentrate. Simple.

The final benefit I’ll identify is economic. Many organizations operate acres of floor space, all lit with tubes. Take half of the tubes away and the lighting bill is cut in half.

I’ve given several good reasons for reducing the light levels of programming shops. Some readers may notice there’s an argument I’ve not made. There’s plenty of anecdotal accounts of full spectrum lighting improving the performance of schoolchildren. The trouble with this stuff is that it is very anecdotal - the more I look for substantial original research on this, the vaguer and more associated with interested parties it becomes. The remaining research doesn’t do a good job of excluding multiple sources of variance, or “confounds” as they’re called in the jargon. The report Full-spectrum fluorescent lighting: a review of its effects on physiology and health by McColl and Veitch, funded by the National Research Council of Canada does a good job of exploring the lack of solid evidence in this area. So although you might choose to experiment with full spectrum lighting, bear in mind that it is more expensive, and the evidence for its benefits is not yet strong.

Experimenting with reduced light levels is a different matter. It has a negative cost. So try it. Persuade your boss to take away some of the tubes (don’t make things too dark - make sure illumination is enough to be safe for example), and see if your working environment becomes cosier, more private, quieter and more relaxing. See if your ability to concentrate improves, and over a release cycle see if that has any effect on the accuracy of your estimates, the necessary sufficiency of your code, and your bug reports.