Yesterday I checked Innerfidelity's frequency response measurement of the UE6000 in Passive mode and compared it to the Sennheiser HD800. Compared to the HD800, the UE6000's treble is down** the following amounts: -6 db at 3khz, -7 db at 4 to 5 khz, -11 db at 6 to 7 khz, and -9 db at 8 khz. The HD800 has a fairly strong treble that's not excessive, so at 6 key frequencies from 3 to 8 khz, the UE6000 is down an average of 8.5 db. Is that a lot? Yes, but it might be even worse. If you have a sample that varies from spec by 3 db (which is probably generous for a product with an iffy QC record), if you add a variance of 3 db to that you're then down an average of 11.5 db, which is very bad. If you subtract a variance of 3 db from that then you're only down an average of 5.5 db, which isn't quite as bad. And that's why it's important to have a tight spec, plus not be far off from a reasonable high fidelity standard, so the variances don't kill your sound. And it probably explains most of the review differences.
**For example, if the HD800's output at 5 khz is 6 db lower than its average midrange output and the UE6000's output at 5 khz is 9 db lower than its average midrange output, then the UE6000's output at 5 khz is down 3 db relative to the HD800.
Last thought: I don't know where Innerfidelity's UE6000 sample measurement falls among the samples being sold now. If their sample had treble output greater than the average sample, then the lower-than-average samples would be down twice as much as I speculated on. For example, you'll see in the All Measurements PDF that some of the famous brands which have measurements of two or more samples sometimes vary greatly in the measured responses.