On unique identifiers: Jimbo replies
From: Jimmy Wales
Date: July 6, 2005 10:03:04 PM GMT+03:00
To: Ulla-Maaria Mutanen
Subject: Re: WIKIPEDIA LTIN
Hello and nice to hear from you again. I’m sorry I’m so slow in
writing but I am always a week behind plus I took a week’s vacation in
France. Back at work now, though.
A problem with human-readable ids is that they will give rise to an
enormous number of problems with trademarks and special words, thus
causing a huge number of useless legal complaints and fights that would
add significantly to the cost of managing the system, and thus to the
cost of end-users. That would defeat the purpose, I think.
I was thinking more in terms of a randomly-assigned 128-bit identifier.
128 bits can be represented in 8 bytes, and is a huge; Probably
using something that looks familiar to people is a good idea, though,
maybe something like hexadecimal.
What might be nice would be to choose a format that is somehow
compatible to existing barcode formats. I’m not sure how to express
what I mean.
What I envision is a system like this, at the core:
1. I have made a product or service. Anyway, it is a
sort, and I wish to sell or trade it.
2. I go to a website and enter my information about the thing. Some
parts of this are permanent and can’t be changed. (The name of it, what
it is.); Other parts could be changed later (my address, my price for
; I am given a code, a code that doesn’t look too scary even though it
isn’t a human-made word (to avoid trademark problems); Maybe it looks
Instantly, then, my product can in theory be placed into all kinds
of different databases either automatically or whatever.
My thinking is that the ecosystems which may build on the identifiers
should be kept separate from the identifiers themselves. Let some
people figure out how to use Ebay-style points systems, let others
figure out how to use a Wiki to describe things in a community way, let
Amazon figure out how to make a transaction system so that sellers can
contact buyers through Amazon, etc.
The advantage is that the product identifiers themselves are not
controlled by Amazon, etc. They are universal and widespread enough (we
hope) that Amazon will be forced to use them. (They now use ASIN, their
own system, but they also use ISBN because the market forces them to do so.)