The aim of tiers is to always have some standard in which a Pokemon will be useable. This means tiers are split in two ways: firstly by centralization, and secondly by useage. Jump back for a moment to RBY, and say you had absolutely no restrictions on what Pokemon could be used. Everyone would always use Mewtwo and 5 Mewtwo counters. It's clear that the presence of Mewtwo causes every single match to revolve around Mewtwo in some respect - either Mewtwo itself, or something that deals with Mewtwo. So, the solution is to ban Mewtwo from the current standard. Mewtwo ends up in the Uber tier.
So, now you no longer have a standard that revolves entirely around one Pokemon. Instead, it revolves around 15 or 16 Pokemon, and the remaining 70 odd don't see any use. Banning one of those 16 doesn't really help. When you banned Mewtwo, the metagame went from revolving around 1 Pokemon to revolving around 16. If you ban one of these 16, it doesn't cause more Pokemon to be viable - you'll just have 15 in use, because the disparity in power between these 15 and the next group is sufficiently large. There's no point in individual Mewtwo style bans, so you just ban by useage - for example, all Pokemon that appear in 1 in 20 matches or more, are banned from the current format. So now, we have an Ubers (with Mewtwo), an "Over-used" with the 16 most-used Pokemon, and an "Under-used" with the rest.
We'll now see a pretty similar problem as before. Dragonite wasn't one of those Pokemon used more than in 1 in 20 teams. However, in this new "Under-used", it is now so powerful in comparison to everything left, teams revolve entirely around bringing Dragonite out, or bringing Dragonite counters out. Dragonite is this banned from the Under-used tiers, and ends up in Border-line.
We've still not solved the problem though, because this new Under-used revolves around only another new set of 15 or 16 Pokemon, and the remaining 40 not in Ubers, OU, BL or in the top of UU still don't see any use and aren't really viable choices. So, we have to repeat the process of banning the Pokemon which are most used in UU. So, take out the Pokemon which are used in more than 1 in 20 UU teams, and you have yourself a NU tier. Again, as before, there may be some Pokemon (or even a few) that centralizes this tier, and they'll be removed to Borderline 2. This is how the tiers develop:
Ubers
OU
BL1
UU
BL2
RU
BL3
NU
...and so on. The names are a little confusing because they evolved organically as terms. A better convention might be something like:
Ban List 1
Tier 1 Useage
Ban List 2
Tier 2 Useage
Ban List 3
Tier 3 Useage
Ban List 4
Tier 4 Useage
...and so on.
The result is that no matter which Pokemon is your favourite, you should be able to find some metagame where they can be viable. You love Walrein, but everyone else is using Garchomp and stomping all over you? No problem! You can play UU, where Walrein can Hailstall everything to the bitter end. You love Beautifly, but everyone else is using Walrein and freezing your dreams? No problem! You can play NU, where Beautifly can be a surprisingly good sweeper/sleeper. That's ultimately what tiers are for. You don't have to pick from a specific list of Pokemon at all. It's just a way of "moderating" fights - preventing people from bringing guns to a knife-fight. If you play Over-used, it's pretty no holds barred except for the super-legendaries being removed, so you don't really have to worry about your choices apart from whether they're likely to net you the win or not.
Most of the bitching from tiers comes from the fact that while it's quite easy to tell whether a Pokemon is Tier 1 Useage or Tier 2 Useage (we can just measure useage from server stats), it's much more difficult to separate between Tier X Useage and Ban List X. You can't really measure it very easily. It is technically possible to do (ban a Pokemon, then see if the number of Pokemon being used in more than 1 in 20 teams is higher or lower than before. If higher, it wasn't centralizing and can be unbanned, if not, retain the ban) but that can be a very difficult and arduous process because you have to allow some time to allow everything to settle after something is banned, and even then there are useage trends which change over time, and it can be difficult to separate changes in useage trends from changes as a result of a ban.
The result is that most of the argument about whether something should be sent to a Ban List revolves around how personal experience of how powerful that Pokemon is and how much you bear it in mind when designing a team - after all, the more you think about it and build your team about, presumably the more freedom you'd have in team-building were it not there. That invites personal anecdotes rather than statistical data into the process, which with a strong and passionate community is obviously where things are going to get messy.
It gets particularly messy with regard to Tier 1 Useage (Over-used, or OU) and Ban List 1 (Ubers), because of the fact the tiering process was until recently rather more organic and has only become a more structured and (arguably) accurate process lately. Ubers for Gen 1, Gen 2 and arguably Gen 3 was truthfully only a ban-list, occupied by a few Pokemon that would have dominated any larger pool into which they were released. Since Gen 4, Ubers has arguably become a largely balanced metagame in its own right. There are so many super-legendaries that they all balance each other out so to speak. However, people associate the tier title "Ubers" with these super-legendaries such as Mewtwo, and thus "Ubers" remains the top ban list.
As such, at the start of each generation, Smogon isn't doing the process described above: starting with everything unbanned, and then proceeding step by step from there. Instead, certain assumptions are made straight off the bat based on these preconceptions of what the Ubers and Overused tiers should look like. The result is that there are many different ideas of what should lead a Pokemon to be put in either Ubers or Overused. Some people view it from this "Ubers is the tier that contains Mewtwo and the other super-legendaries by default" perspective, the more evolutionary perspective. Others view it in the more procedural manners, which is "Ubers it the tier that contains only those Pokemon which cannot possibly exist in a balanced metagame".
Given Overused is the most played standard, and given there is a need to sort Pokemon into categories, but with insufficiently clear criteria and a heavy basis on personal anecdotes, it's inevitable there will be a bit of turbulence. Personally, I lean more towards the procedural system. For X and Y's metagame, I'd prefer to see a "first 6 months have everything unbanned", and move on from there. However, the evolutionary perspective is not wrong per se, it's just a different way of approaching the issue. Typically, the procedural approach tends to be rather liberal about banlists, aiming for minimalism, and the evolutionary approach tends to build on banlists previously established in past generations, so they expand over time.
Anyway, probably went slightly too into depth there and I'm "retired" from the scene so particular examples may be dated/wrong (no idea what tier Walrein is these days, just guessing), but that's how tiers work and why there's a fuss about them. They're important things to have, but given how complex Pokemon is, they're rather hard to get right.