Indeed but, for those people that have a need to implement this functionality, it saves both band width and processing time, both on the application and the database server. If we were doing this via a dataset returned from the client with say twenty records in, you'd be looking at twenty individual updates, with all the locking and extra processing time this would require. The solution in the article requires only one call to the database and does not need the rest of the data in order to do the resequencing.

I would agree that scalability would be an issue if what was being resequenced was a public list that many people were updating simultaneously but the article is aimed more at business applications, where these are user-maintained lists and, as such, they are unlikely to be updated by more than one person at any given time. From personal experience I can say that this does scale well, indeed far better than a recordset based solution.