Brad DeLong is running some fascinating posts on productivity and technology. A good place to start is this piece, on the data showing rapid declines in the cost of hardware, with no corresponding decline in the cost of software. Brad’s piece is headed “Where Is Moore’s Law for Software?, or, The Mythical Man-Month Strikes Back”
My view is that both hardware and personal computer software have run into diminishing returns. In the case of hardware, this has been offset until recently by Moore’s Law – the biennial doubling of processor speed, memory capacity and so on. In the case of software, the opposite is true if anything. Not only has the marginal benefit of a line of code diminished, but the rising real wages of programmers and the rapacious demands of Microsoft have pushed the price of code up.
Economists profess not to like introspection and anecdotal evidence but we actually rely on it all the time. So I’ll draw on my own. I’ve realised huge increases in productivity from personal computers and I’m always willing to spend money if it will improve my productivity further. But all the really crucial innovations, as far as my ‘pure’ research is concerned were between 1984, when I got my first Mac, and the early 1990s, when email with attachments became generally available for academics. This set of innovations mean that I now routinely work on a shift system with co-authors in the United States. In the morning I get the latest instalment from my co-author, and in the evening I send an amended version back to them. I estimate that this system, along with the prior innovation of word processing, has at least doubled my research productivity and probably quadrupled it (in quantity terms, anyway, but I think there’s also been a quality improvement from the ability to collaborate with colleagues on the other side of the world).
Throughout the 80s and most of the 90s I was always aware of hardware constraints. The processor was never fast enough for Mac OS, there was never enough hard disk space and never enough memory. I remember the relief when, thanks to a price war around 1990, I was able to pick up 20Mb of RAM for only $1000. And that was nothing compared to the time saved when I got my first 20MB hard disk in the 1980s. Today, I’m like the apocryphal millionaire (Getty?) – if you know how much RAM and disk space you have, then you obviously haven’t got enough. I could easily double both at low cost, and upgrade my processor to one twice as fast, but I can’t be bothered doing the necessary paperwork (and it’s still literally paperwork!) or taking my Mac into the store. The only margin on which improved hardware is still contributing to my productivity is that of monitors – bigger, brighter, flatter and multiple, I still can’t get enough. But of course, Moore’s Law doesn’t apply here. It’s taken fifteen years for the standard monitor to go from 13 inches to 21.
It’s a similar story with software. Throughout the 1980s, a new version of a word processor was an event to be eagerly awaited. Now the only reason I bother with upgrades is to maintain file compatibility.
The discussion above leaves out the Web and blogging, which I’ll talk about in a later post. For the moment, I’ll just observe that, as far as my research in economic theory is concerned the impact of the Web on my productivity has been modest or, in its capacity as a method of wasting time, negative. As for blogging ….