Of course you neglect to mention that it is pretty much only FOREIGN companies that display this tendency. It was demonstrated in post 250 in the "old thread".
From that post we have the following;
Finally, we have ARCSIGHT INC
Sure enough, it IS a US corporation. And sure enough there ARE records with the same start date. HOWEVER, if you look at the applications they filed for the TEMPE office, they have one application for 1 SOFTWARE ENGINEER each on the same date and TWO applications for 5 SOFTWARE ENGINEERS with the same date as the other applications. It becomes readily apparent that if they filed even ONE application with only ONE worker and ONE application with FIVE workers all with the same start date, that it is IS common practice to file separate applications with the same start date for the same location, salary, and job title.
You've just proven that your "bad data" theory is a "bad" assumption.
One WOULD wonder why ARCSIGHT is hiring all of those H1-B workers in Sunnyvale and Tempe when there are so many engineers with those skills who can't find work. Are they saying that they couldn't find any software engineers with the necessary skills in Silicon Valley or Tempe, AZ? Right.
And as far as DCM goes, they DID lie and call their positions "programmer analyst" where the work they describe on their site relates to software engineering. They DID file applications for 30 "programmer analyst" not only for their Austin development center but for each and every one of their sales offices thought the US. And they DID file 104 LCA applications resulting in over 3000 H1-B visas where they claim a workforce of 300.
So, here we have demonstrated that it is NOT a database error, it is what was INPUT into the database.
It's wierd, not invalidating, just wierd and something worth keeping in mind when looking at the data.
It's weird alright. I'd say it's something that SHOULD be looked into. We know that it isn't a known error, as the FAQ lists ALL known errors in the database. Duplicate records is NOT one of those errors, as we know from post 261 in the old thread...