Any recommendation to stick with 2TB drives because they're magically more reliable, is overly simplistic and not really helpful. There are some reliability differences across manufacturers, models or even batches of a specific model. But these issues have very little to do with something as trivial as saying that capacity X is reliable and capacity Y is not. I've had no issues with various 3TB and 4TB models, mostly Western Digital Green (for desktop) and RE2 and Red (for server/array) models for me, over the last few years.
One thing that may affect reliability, which we have discussed here many times before, is that the cheaper desktop hard drives are not designed to run 7x24 in storage arrays. So you're taking a risk if you do that. I always use drive models designed for servers and RAID arrays, in my servers and arrays. That's true whether I pick Seagate or Western Digital. I don't run enough capacity to be able to amortize the higher failure rate of desktop-designed drives vs. their lower cost compared to server drives; also, my time is too valuable to intentionally do things that are more likely to cause me trouble down the road.
Also as we have discussed here many times, I don't recommend basing overly much of your choice on a specific model or capacity of hard drive or enclosure, anyway, especially if it's somewhat new and not much operational info has had time to accumulate. Instead, you should have a robust backup & recovery approach that plans for and tolerates the failure of components without exposing you to large risks of data loss. If you get a single 4-bay NAS box regardless of who made it or what kind of drives you put in it, how are you going to back it up? Or is it supposed to be the backup for your workstation internals? If the latter, that would sound like you have only 2 copies. Either way, what if the NAS box itself (not the drives in it) fails? Unless you have a 2nd identical box you can't just remove the drives and stick them in something else and have it work. Boxes like this change all the time; unless you bought 2 identical ones at the same time, you can't guarantee your ability to buy a replacement box in the future that will be plug-in compatible with the dead one. So you'd potentially lose the entire contents of the storage array and have to re-copy it all again from another source, even if the manufacturer replaced a dead box with another comparable (but not identical) model.
I recommend thinking through things like this before creating a giant basket and putting a lot of eggs into it. Having a single loaded storage array is a risk point in itself. Having only two copies of data is also a risk. Combining both of those factors together is a multiplied risk. 3 copies with 1 copy powered off & disconnected (ideally rotated offsite to a secure location) is the minimum to consider as even a basic backup approach. If one of those 3 copies is the storage array, and it goes down, then you're sitting with only 2 copies. Probably both of them will be in the same place and live connected to each other for the duration of time it takes you to recover the storage array. How long will that be? During that window of time you're now at an increased risk of losing everything if something clobbers the 2 copies that are live and in the same place. That's why I run more than 3 copies, and never have a single brand/model of storage array -- I always buy them in 2's or 3's so they can back each other up in the event one of them fails.
Back to hard drive models, for a good snapshot of info on reliability, see this article from cloud storage vendor Backblaze.
http://blog.backblaze.com/2014/01/21/wh ... uld-i-buy/
Note that they don't buy enterprise-grade drives designed for servers & RAID arrays. So I think at least in the case of Seagate models, their data supports the argument that using desktop drive models in arrays increases the risk of failure in 7x24 arrays. Look at the failure curve plotted over 36 months of Seagate vs. Western Digital vs. Hitachi (now owned by WD). These are all consumer grade desktop drives, and clearly there is a manufacturer difference at work here, with Seagate not doing nearly so well due in large part to certain specific models.
Out of hundreds of units in use in their storage arrays, Backblaze data shows that one model of Seagate Barracuda 1.5TB suffered a ~25% failure rate, while another model of Seagate 1.5TB suffered a ~10% failure rate. A third Seagate Green 1.5TB model with low count of units suffered a whopping 120% failure rate. So clearly these 1.5TB technology-based drives from Seagate had issues. (This flies in the face of a simplistic recommendation that anything over 2TB is an unsafe choice, because many of the Backblaze models they report on are far more reliable than these smaller 1.5TB models.) A Seagate 3TB model also did suffer a ~10% failure rate. But generally I feel the issue is partly specific Seagate models, but more generally it's likely related to using desktop Seagate drives in 7x24 arrays. You can read the article for the rest.
Obviously armed with info like this, one could make some generalizations about what to buy or not buy. But it has little to do with capacity, and more to do with manufacturer or model-specific issues...