Kuhn's idea

Philip Greenspun's Homepage : Philip Greenspun's Homepage Discussion Forums : Ask Philip : One Thread
Notify me of new responses
I brought up the point of progress in Computer Science being made more
by books rather than conference papers and my friend argued that more
progress is made by the software that was written/distributed rather
than books. I thought that was wrong because books are more general
and more accessible to more people, but it's hard to say what
contributed more to the success of ACS, the fact that it was
open-source or the fact that there was a book describing architecture
in which the ACS fit into well. But then I realized that the book
would not have made as much of an impact if a lot of software had not
previously been written & distributed, such as web servers, operating
systems, etc. So I was just wondering what you thought of that.

Was there an outage? Looks like some recent questions were lost..

-- Neal Sidhwaney, March 3, 2003

Answers

As far as computer science theory goes it is probably the books that generate the largest shifts in thinking, e.g., Preparata and Shamos's Computational Geometry. But for practice it is definitely the software, as your friend suggests. The Macintosh, for example, had a huge effect on user interface design. It is true that the principles were nicely codified in a book but the thing that stuck in most programmers' minds was the experience of using a Mac. Many of the Xerox PARC ideas got spread most effectively when they exported physical machines, networks, and laser printers to other computer science research labs (the MIT AI Lab, for example, had a big laser printer and some workstations from PARC).

As far as the important ideas of the old ArsDigita Community System ("ACS", a toolkit for building Internet applications) are concerned, in some ways the existence of a popular ACS-backed site such as photo.net will have more long-term influence than the software per se. Developers at Microsoft, for example, who have used ACS will incorporate the good user experience ideas into the products that they build.

You ask a separate question, Neal, from what Kuhn was interested in. Kuhn was interested in the development of knowledge rather than the popularity of an intermediate idea. You ask, however, about what contributed to the success of ArsDigita Community System, a specific product circa 1999. The book and the related English-language documents were definitely critical to its successs. People don't have the time, patience, or risk-tolerance to wade into the source code for a complex piece of software or to adopt a piece of multi-user software and try it out. By contrast to read a book or a document describing, say, the ticket tracking system in ACS, is a low-investment, low-risk activity. Unless you're an Oracle, IBM, or Microsoft it would be virtually impossible to get someone to take a chance on a new software system without having enticed them into it gently.

Which I guess feeds back to the initial question of whether books or software are more important. With enterprise software, if you don't have books or an overwhelmingly strong market position, nobody is going to try out the software even if it is free. So perhaps books are essential to getting ideas across in this area. With desktop software that an individual can download and try out without risking more than one person's time, it would seem that the software alone can spread the ideas within.

Regarding the missing questions... I just deleted some of the questions that I thought had gotten their 15 minutes of fame.

-- Philip Greenspun, March 3, 2003


You ask a separate question, Neal, from what Kuhn was interested in. Kuhn was interested in the development of knowledge rather than the popularity of an intermediate idea. You ask, however, about what contributed to the success of ArsDigita Community System, a specific product circa 1999.

I guess what I meant by success of ACS was more the success of ideas behind ACS, like db-backed web sites having a community-aspect using bboards, letting people comment on all objects(static pages, pictures, etc.), or scalability through connecting to the database intelligently, and not widespread use of the product itself. This seems to be the development of knowledge. I can definitely see how a running example of the ideas would be more useful than just the distribution of the software.

-- Neal Sidhwaney, March 3, 2003