I seen this post as well, but never seen any good hard data behind it. Were any parses done to confirm this? Sorry to throw your quote into doubt, but after years of XI there have been many claims like this that are just flat out wrong.
EDIT: I haven't noticed a threshold like potency increase with INT on THM, but I will check it out tomorrow on some 50 drubbers. It just doesn't make sense that SE would make a threshold based system like this, as it would clearly render most stats useless except the one you are going all out for.
EDIT2: Went ahead and went to the 50 drubbers, took off some INT to get under the so called "340 threshold" in that guy's post. The closest I could get was 336 INT, which I did a few Thunders with, then I put a ring back on to put me at 343 INT, then I tried again with my usual 363 INT:
336INT: 527, 516, 496, 492, 521, 516, Avg: 511.3
343INT: 530, 528, 508, 548, 527, 546, Avg: 531.1
363INT: 532, 534, 532, 548, 549, 556, Avg: 541.8
There aren't nearly enough nuke samples there to come up with a concrete answer (out of 6 nukes with each setup, the jitter could easily throw the average a bit high or low), but I think we can easily call shenanigans on this theory as far as INT goes.