Archive for December, 2014

All Hammers are Terrible

December 29, 2014

“All computers are terrible.”

On social media, this turn of phrase is almost a meme. The sentiment is reasonable at some level—computers are complicated, buggy, and insecure. But the general public is largely indifferent, or least, they accept the situation for what it is: Computer freezes? turn it on then off again. Smart phone insecure? Well, the convenience of digital wallets outweighs the risk. My parents, who are not computer scientists, never say that all computers are terrible.

Rather, this is a complaint from within, by computer scientists themselves.  What drives this thinking? (While not specifically addressing the cause of the sentiment, this blog post provides a nice counterpoint.) No immediately obvious hypotheses seem particularly explanatory:

  • Ludditism: perhaps despite (or because of) our work in technology, computer scientists abhor it. But I find this explanation wanting, particularly because the focus of contempt is specifically about computers, and not about technological advancement generally.
  • Pessimism: perhaps there is a correlation between computer scientists and pessimists. Faced with the challenge of building good systems, they focus on the flaws rather than the progress. But it does not adequately explain the focus of the pessimism on computers rather than life in general—I don’t see posts stating that “all relationships are terrible”.
  • Realism: perhaps computers are terrible. Computer science history is measured in decades, and the approach to building systems is immature compared to other engineering disciplines. I tore down my old garage this summer. The city required a blueprint, erosion plan, a drainage plan, and a few hundred dollars for me to do so. They inspected the work afterwards. Contrast this with software: how much oversight does writing security-critical code get? Just about anyone can throw code up that purports to solve a problem. Still, given that programs are among the most complex artifacts built by humans, we’re doing pretty well.
  • Narcissism: finally, perhaps a tinge of narcissism drives these comments, something along the lines of, “where others have failed, I will succeed.” It is given as a tongue-in-cheek put-down regarding the work of fellow professionals. The motivation is there: if we thought that the tools of computer science were “good enough”, then the world wouldn’t need new programming languages, tools, or libraries. On the other hand, these kind of complaints are not just about directly competing work, and there will always be new inventions to make.

I tend not to hear the same kind of complaints from members of other professions. Yes, we all complain about our jobs, but not necessarily about our tools. Carpenters don’t say, “All hammers are terrible,” even if they complain about their company, their coworkers, the architect, the inspectors. Doctors don’t complain about X-ray machines, MRIs, or scalpels (but they might complain about their patients, insurance, and administration). Farmers don’t complain about the design of combines and tractors. Now, they might complain about a specific instance—“My tractor is terrible; it breaks down all the time.” But not about tractors, in general.

Other professions largely take their tools as a given. Indeed, most practitioners know so little about how the tools they use are built, it is difficult to critique them. A carpenter does not know how a skill saw is built and a doctor does not know how an X-rays are taken. Just whether they work or not and if they are useful or not.

Computer science is unique insofar as the tool makers are so closely connected with the tool users. There’s another glib quip in computer science: “patches welcome” meaning that if you (the user) thinks a tool should be improved, then it is your responsibility to make those improvements yourself. In other fields, the saying does not exist; it would be laughable! A taxi driver can’t be expected to invent a more fuel-efficient automobile; an architect can’t be expected to build better modeling software.

Computer science is unique in this respect, and it has philosophical, psychological, and sociological implications. Imagine, if you will, you are a carpenter, and you have some understanding of how the tools that you have are designed and built. Your skill saw is right-handed, but you are left-handed; with a few modifications, you realize its design can be abstracted, to accommodate either handedness. When your drill dies, instead of simply replacing it, you open it up, find the faulty circuit, and realize there’s a weakness in the wiring that allows it to overheat. When each of your tools fails you in some way, you uncover some flaw and realize they could be improved. Finally, you come to expect failure from every tool you might ever use: “All hammers are terrible.”

We all know that technology—whether we understand how it is built or not—has flaws. We know tradeoffs are made between cost, generality, performance, and reliability. There are more and less reliable, performant, and costly cars, saws, X-ray machines.

The general public certainly has opinions and a sense of the tradeoffs between the tools they use, e.g., Windows vs. Mac OS, Android vs. iOS. But the sentiment is that these are tools, with flaws and lifespans, just like a cars, appliances, and power tools.

I don’t have a strong opinion about whether computer scientists should complain or not about their tools, except, obviously, that general complaints don’t effect much change. My point is more philosophical: we find ourselves in an interesting profession, where we can open our very tools up, peer inside, and change them.

It’s a fascinating prospect.