Haswell and IB overclocking has been severely limited because Intel uses cheap and lousy TIM instead of fluxless solder. Just by delidding and applying a high quality thermal paste you can improve temps by as much as 30+ C. Its so substantial its really not funny.
Just a note though, the thermal paste that Intel use are very high quality... yet they handicapped the performance of OCing by putting a unnecessary thick seal which cos the heat to conduct through the IHS very slowly. It is a marketing means to deter enthusiast from maximizing their "cheap" CPU. As you know Intel is pissing its customer off by removing some of the essential function for the best performance, like TSX for unlocked CPUs and restricted overclocking for locked one.
The new Intel LGA 2011 IVB like 4960X, uses fluxless solder and this further affirm the conspiracy for the enthusiast to pay more for the "higher-end" CPU.
I am very confident Intel is capable of producing much better CPUs than this junk Haswell, its just that they are not putting in effort to make it better thanks to the lack of competition... The next Haswell refresh I'm pretty sure that it will have even better performance efficiency (improved IPC performance at lower wattage and heat), tempting ppl to change to this refresh series from their current haswell, IVB or even SB. Yet at the same time crippling certain function like they did now..
This is done to prepare for the release of their real performance monster. The Intel Haswell E. (I already find it suspicious why Haswell can maintain a 130W TDP despite having 2 additional cores and running on a power hungry FIVR, the only reason to explain is that they will use the improved haswell refresh to run this monster so it wont overcook the chip.)