Thursday, 16 July 2009

Discovery vs creation

One of the biggest differences between mathematics and programming is the question of authorship. Advances in maths are typically described as discoveries, whereas new software is developed, created or invented. Though programming and mathematics employ similar notations, the uses of these notations are governed by strikingly dissimilar discourses.

Even Kurt Gödel, who's incompleteness theorems are perhaps the most well-known examples of the limitations of mathematics, is widely regarded as a Platonist. He, like many mathematicians, regarded mathematics as more real than the physical world. For a Platonist, theorems are timeless and eternal. Mathematicians' role is to discover and document them as purely as possible. Paul Erdős expressed this sentiment by imagining that the most beautiful proofs came from a book written by God.

On the other hand, few would claim that Linux existed before Linus Torvalds started writing it in 1991. Even a software engineering concept like structured programming is usually described as being founded by Edsger Dijkstra, even though the mathematical theorem that underpins the movement could be said to have been discovered (by Corrado Böhm and Giuseppe Jacopini).

Some mathematicians do leave room for authorship in their understanding of their profession. Leopold Kronecker once said that God made the integers, all else is the work of man. Bertrand Russell went further and said that integers were also created by man - or at least they could be constructed using mathematical logic.

The defining characteristic of authorship (as opposed to invention) is that the subjectivity of the author is imprinted on the work. One example of this in mathematics is the calculus. Isaac Newton and Gottfried Leibniz both discovered the calculus, but they approached it in different ways. I would argue that their divergent expressions of the same idea are best understood through the lens of authorship, especially given the importance Leibniz placed on notation and presenting his thoughts for human understanding.

But by and large, mathematicians are better described by Roland Barthes' account of tellers of tales before modern authorship was invented:
In ethnographic societies the responsibility for a narrative is never assumed by a person but by a mediator, shaman or relator whose ‘performance’ — the mastery of the narrative code —may possibly be admired but never his ‘genius’. The author is a modern figure.
Mathematicians attach their names to their work, but more in the spirit of explorers naming newly discovered peaks than authors cultivating writing credits. It is this emphasis on discovery rather than creation that most clearly differentiates mathematical practice from programming and which means that a purely mathematical education is not sufficient to understand software development.

Wednesday, 1 July 2009

The professional is political

I wondered once whether my occupation influences my politics. Then I thought, of course it does. Construction workers are more likely to be unionists, small business owners are more likely to be economically conservative and Barack Obama's chief of staff is more likely to be a Democrat. It's hard to make definite predictions (except possibly the last one), but what you do for eight hours a weekday has to influence your view of the world.

It's not immediately obvious what political stereotype to apply to software developers. One the one hand, computer systems are tightly controlled, deterministic universes where users can only venture if they provide the correct password (which the programmer has decreed shall contain no less than three non-alphanumeric characters).

This suggests that programmers might have sympathy for centrally planned economies. Citizens' input will correctly validate or they will be re-educated!

On the other hand, a lot of our time as software designers is dedicated to preserving flexibility. We use factory methods and interfaces to give ourselves the freedom to change which class we wish to instantiate. We attempt to compose methods so that they can be re-used in other contexts. Nathaniel Borenstein captured this attitude perfectly:
It should be noted that no ethically-trained software engineer would ever consent to write a DestroyBaghdad procedure. Basic professional ethics would instead require him to write a DestroyCity procedure, to which Baghdad could be given as a parameter.
Some use C++ because they don't want some garbage-collecting nanny state managing their memory for them. I'm sure that Margaret Thatcher would have agreed with this sentiment had she studied programming rather than chemistry - though her declaration that if you want to cut your own throat, don't come to me for a bandage does suspiciously like she's warning a junior developer away from pointer arithmetic.

The common theme of these examples is that software developers are constantly struggling to capture logic at the appropriate place in their code. This is the key point I take from software to my own political opinions. When it comes right down to it, most of the political ideas and philosophies that I dislike are operating on the wrong level of abstraction.

Centralised steel quotas are a bad idea because production decisions are best made on a local level, not in Beijing. Internet censorship is problematic because an individual is best placed to decide what they do not wish to view. A single point of control is not appropriate for these examples.

However, some political decisions cannot be left to the individual. Controlling greenhouse gas emissions is a good example of an issue that needs to be managed centrally to avoid a tragedy of the commons.

My instincts have always been libertarian. But my experiences as a software developer have taught me that there is no hard and fast rule for what level of abstraction decisions should be made at. As much as I'd like individuals to be given complete control over their lives, sometimes individuals simply do not have the necessary perspective to make the best decision.

For example, ethical consumerism is a laudable philosophy, but it does not work unless there is some central agency capable of understanding the consequences of individual purchases who can guide consumers. And government schools are necessary because leaving it to parents to purchase education for their own children will lead to unacceptable inequality.

Considering issues in isolation can also lead to short-sighted decisions. I disapprove of the Californian system of referenda, because of course people will vote for lower taxes and higher spending if they are asked about these issues in isolation. Budgets need to be created from a perspective that allows consideration of all of a government's finances.

To paraphrase Einstein's well-known quote, make your code as simple (and generic) as possible, but no simpler. Give individuals as much liberty as possible, but no more. And when contemplating a political dilemma, consider what level of abstraction the problem would be best addressed at.