dataset or technique?

An interesting trend is clearly noticeable, Sensor data sharing website. There are plenty of websites on internet collect data from numerous sensor networks and published the data online. With the increasing popularity of micro-sensor network there is no wander that many researchers are exploring different facets of sensors network applications. Myriad of research papers have been published throughout this decade to address fictitious or real problems. I am not sure how many of them are truly useful but at least those papers help to some students to graduate and some professors to be tenure. Quote unquote, "Ian F. Akyildiz" this area is too crowded but industry usage is too restrictive. It is difficult to point the real problem that discourages the industry people. Anyway, it was not my concern what I am trying to point out is that there are numerous sensor data sharing websites on internet and many researcher collect data from those websites.

The scenario becomes too fuzzy now. It's like someone is using some others' data for some other purposes. Based on my observation, the results published in this domain are highly sensitive to the dataset. Now if the dataset doesn't make any sense, how come the results! The websites are providing data only nobody cares about the quality! I know it is neither their responsibility nor concern but this is the researchers' concern. In my opinion any researcher can write a proposal about the quality estimation for these data and get some funding from them. To me it is a clear need now. I was reading a paper this morning, it was a survey paper and in that paper the authors argue, the results do not vary from one technique to another technique, rather the results vary from one dataset to another dataset! So what is more important, come up with a new technique or come up with standard dataset?