I am a fan of the culture of open problems, posing them, solving them, passing them around in parties to prod people into action. Two collections of open problems are interesting:
- Andrew McGregor has compiled a bunch of open problems on data stream algorithms, with background and bib notes. Excellent!
- Jiri Matousek has a fantastic tome of open problems on low distortion embeddings, recently revised to also show solved problems since the origins of this document in 2002.