Monday, May 22, 2006

Network effects

I have disabled blog comments for folks not registered with blogger.com. Sorry numismatic readers. Spammers were leaving comments.

R.J. O'hara points out in a blog comment that publishers must underestimate demand, since scholaly numismatic publications go out of print fast and cost even more on the used market than they cost new.

Certain numismatic books will never be popular. No matter how many copies were printed of SNG Newcastle, no matter how good Andrew Meadows' writing turns out to be, it was too many copies.

Numismatic references, the kind with a number-per-coin, are like dictionaries to the language of numismatics. A few books capture the discourse and become popular. These books are necessary to understand the global coin 'conversation.' There is a network effect — once a few people start using a numbering system it becomes more useful to use that system than to use a better but less popular system.

When someone asks me to identify a coin the reference books I consult are ones 1) likely to have the coin and 2) that the questioner is likely to have. For a book to become popular it has to be be strong in #1 (findability), to compensate for being new and having no #2 (popularity).

That's how I know SNG Newcastle will never be a hit — it has 1000 coins from all over the Greek world. If it had 1000 coins from Thessaly I would want to consult it. 1000 random Greek coins makes the odds of finding a specific coin too low to justify walking over to the shelf and picking it up. (Luckily SNG Newcastle is part of the great sylloge-nummorum-graecorum.org/ so people will still use it, and scholars will still quote it. It just won't sell many copies.)

A book like Roman Provincial Coinage I includes every provincial coin from 44 BC to 69 AD. This makes the findability 100%. It is the only book for the period with 100% findability. So of course it became popular among numismatic catalogers and writers. Once a numbering system is adopted by other authors suddenly everyone else needs the book to look up the numbers and see what they mean, ensuring continued popularity. A book like that can never have enough copies printed, no matter how big the print run.

1 comment:

Robert J. said...

Right on target! That's the best short description of the phenomenon I've read. (You should work it up as a little editorial for the Celator perhaps.)

This is why I got the two very nice ANS "Ancient Coins in North American Collections" volumes that cover the Washington Univ. and Wheaton College collections for less than $10/each: both are broad, general collections, great for their institutions but of little reference value to the population at large. But the publishers should recognize that something like RPC is destined to be a universal reference, not like these two, and should increase the print run (and decrease the cost) accordingly.

Your note also gives good advice to would-be authors/publishers: if you want to be cited, create a through and simple numbering system that incorporates most everything known. The only exceptions are things like SNG Kayhan and the ANS Rosen collection, which are not exhaustive, but cover a lot of material that just hasn't been widely published before. They are sufficiently comprehensive that they are widely cited, even though they contain a lot of uncertain attributions.

(Shame about the spamming. I think Blogger has an authentication feature you can turn on, requiring people to enter a code shown as a graphic. Perhaps that would work?)

Bob O'Hara