Sloan/MIT uses Oracle - they've got a site license so switching to Postgres would actually cost them money in terms of setting up administrative stuff and all that, plus Oracle works great for them so there's no motivation for them to switch.
Bug Ben for details ... the work will be done in the OpenACS framework, so porting over to Postgres won't be hard. And I'm sure the folks doing the work will try to use vanilla SQL92 constructs wherever possible and avoid Oracle-isms whenever they can.
Porting to another DB, while tedious and timeconsuming, requires just a fraction of the effort that fresh implementation does. A positive way of looking is to remember that this effort will be covering 100% of the time required to together the Oracle version, and probably on the order of 90% of the time required to do the PostgreSQL version.
Most of the work is initial implementation, and a large proportion of the queries can be shared directly will require little work to rewrite.
