We do acknowledge that the presented idea of a search engine does not follow the modern concept that an application should get better with every usage of it; ideally, we should track usage scenarios and make predictions based on those observations, thus finding the tweets the user is looking for. However, we would like to point out that Twitter search does not abide by the same laws as the informative search that has been a research highlight until recently; since there are only 140 symbols, in our opinion, the future of Twitter search lies in identifying the existing trends in the society rather than giving particular tweets that solely match a query by textual features. So, every query, in principle, should be matched with some trend behind it, be it the recent situation in the East or plans for the weekend.
Besides, it should be possible to be able to react to queries like ‘funny tweets’: that kind, obviously, is not intended for processing under the assumptions of the bag-of-words model.
As for the struggles with normalization and acronyms, as well as absense of implementation of word sense disambiguation, prompting of possible intended queries, detecting of discussions on the same topic over the time even with modifying hashtags, and other features that need to be implemented in any major large-scale search engine, we do not see those flaws as conceptual ones – that can simply be treated by going on with the development of the application; it is the overall direction and spirit of work that we think does not correspond to the current needs of search engine technologies for microblogs.

А-ня-ня, как же это приятно, говорить, что задание не сделано не потому что не сделано, а потому что у меня Свой Взгляд ))