To pick a number of comments in this thread:
Submission of records is the basis for all Biological Recording Schemes, requiring a Who (name of recorder), What (species/taxa recorded), Where (site name and grid reference) and When (date and ideally time) for each, plus further information for certain confusion or rare types.
The vast majority of records will be assessed by county recorders and (as appropriate) the local/national rarity committees who would have not seen the bird, etc concerned.
Most records submitted to BirdTrack are accessible to county bird recorders - I receive 35,000 each year for W Mids county alone! Others are submitted directly via record cards, letters, telephone calls, texts, face-to-face, e-mails, Word/Excel attachments, site annual reports, etc, which gives some sense of the size of the workload involved for just one small county. Whilst all this data takes time to assess, assimilate and write meaningful species accounts, it is only part of the issue. Records are not all received promptly e.g. I am still waiting for an annual report for a notable W Mids site for 2011, hence the delay in publishing West Midland Bird Club Annual Reports. The alternative, publish the report without the records, which to my mind somewhat defeats the object.
Without doubt many records are lost because no details are submitted at all, which can undervalue any assessment of a site, species, etc. I do trawl some websites, blogs, etc for bird records, but I certainly do not have the time to actively search for them all, so please do not presume that because records are on the Internet that they will be found by the respective county recorder, etc!
For records of description species submitted or found without supporting information, I contact the recorder if possible for further details, which sadly are still not always forthcoming. If I can find any supporting information e.g. recorded independently by different people on the same or different days or a photograph on a website, the record would be accepted, unless it involves a BBRC species which of course requires their consideration.
Furthermore, if there have been other substantiated records of a species locally e.g. Waxwing in an irruption year, they would tend to be entered as "reports of". On the subject of Marsh or Willow Tit, records without further details would also be recorded as reports of the stated species. Records involving distinctive species e.g. Red Kite where the observer says that they have seen them before, but provides no description, I tend to accept, but otherwise they would also be "reports of". Of course, this relies on trusting the observer and is very subjective, but I try to give the benefit of the doubt and strike a balance. The alternative is that none of these records are incorporated and therefore they all become lost.
If anyone can suggest alternative scientifically sound approaches, please let me know.
Cheers.
Kevin