"Data driven", but signals above 3 sigma still get a lot of interest, and many come and go. With little interest in explaining them. The amount of statistics and data to get from 3 to 5 is quite larger.
But the only thing a bit worry some here is the complete lack of interest in low energies, anything with a realistic economic impact will be found here.
You are right, there is quite a lot of "cuts" done one the data to get rid of noise, and they do other things to search possibilities in the noise.
Neutrino experiments and that anti proton decelerator may be more interesting as many things about HEP is engineering porn!

And much less, "oh shit! we don't know what to expect from this".
Not that many experiments that have studied the neutrino oscillation problem after it was solved.
Good science is to do it all over again to make sure.