A better stemmer does sound attractive. As far as whether this justifies the switch, I have two main concerns to weigh it against: frequency of updates and long-term support. I know we're waiting on Solr 1.4 for at least one feature (spellchecker reindex on optimize), and it would be unfortunate if using a different build significantly delayed our ability to take advantage of the new Solr version or future fixes/enhancements. It could also be a problem in the future if Solr moves on and the LucidImagination build dies -- we could end up becoming dependent on a feature set that is no longer part of active Solr development. I'm not sure how likely it is that either of these things will become problems -- I'm new to Solr and unfamiliar with LucidImagination -- but I think they're considerations worth keeping in mind when making the choice.