It's been long understood that scientific conjectures and hypotheses must be tested independently by people other than the ones that developed the ideas. Thus, Einstein didn't really much bother with experimental confirmation; that was the job of all the other physicists who (quite properly) didn't accept his ideas and were trying to disprove them. Real science does require verification, of course, but there's no reason to insist that it be done by the people who do the theoretical work. Also, there are known problems with people trying to experimentally verify their own hypotheses, which is why we so often read calls for independent testing.
So what's new about all this today? It sounds like Science As Usual to me. A lot of the hypotheses will never be tested, but that just means that they'll never graduate to the class of "theory".
A parallel that I've found instructive: In the publishing industry, it's well understood that proofreading must be done (if it's done at all ;-) by someone other than the author. It's difficult to proofread your own stuff, because you tend to read what you know should be there, not what actually is. I've seen this myself, with people pointing out typos in things I've put online that I know I proofread. I generally just fix the error, and send them a "Thanks" message, then go about what I was doing. Similar comments probably apply any time you're trying to actually get something right in any subject area.
One might be tempted to make the extreme suggestion that people shouldn't bother checking their own work. Just send it to an independent checker, perhaps someone who is willing to send you their work for checking. Send it to several such checkers, who have an understanding that you'll do the same for them. This way, people can concentrate on producing stuff that they're good at, and pay for it by spending time similarly checking other people's work that might not be so close and familiar.
I've seen evidence that this has sorta happened in a few fields. The idea is that you keep all the stuff you're working on online, in a semi-hidden place that your colleagues know how to access, but which isn't really "public". You might send email out occasionally asking them to read through a new document that you're putting together. This sort of setup happens a lot in software development, typically as online repositories clearly labelled as "development" to warn away non-technical "users". A mailing list or blog helps get people together who are willing to download and test new versions and send in bug reports. When you get enough colleages saying it seems to be working, you announce a new public release. This is not really very different from the old scientific concept of independent verification.