Scrum for short projects

Jeff Sutherland has used 1 week sprints in his Shock Therapy approach. I have used them also though in a different context. See this thread:

http://www.scrum.org/Forums/aft/449

A significant factor in determining Sprint length is the ability of the Product Owner to exert pull and to release value.

Let's remember also that in Scrum there is no notion of a "project" of a given length. Rather there is a "product", the backlog of which can be estimated at any given time.

Hmmm... why dont you have a team backlog.... and all the different PBIs from different Product backlogs(which you have mentioned as Projects) be taken to be worked on next. That way your team could work on different products for short durations without having to disrupt the place(Backlog) which they have to look at for more work.

Also if you have to work on different applications for shorter durations it would be best to go on 1 week cycles that way it is easier to get the work out there and it is only 1 week turnaround.

All Scrum says is the max length of the Sprint should be 4 weeks.

As a PO, I've completed numerous projects in short Sprints, some of them broken down into phases / short releases.

One notable one was 6 weeks (3 Sprints), where 1 team saved a Pharma organization $4Million, reduced safety inspections, saved lab space, helped eliminate a $800K renovation project and improved workforce efficiency by 60%

@Cynthia --
I've worked in a similar setup at the Program level. In such a scenario, I agree that a 1-week Sprint be considered and where some "mini projects" are completed in their entirety.

So, to Fredrik's question --
See what makes sense to the organization and the team.