People in this thread have commented on several flaws or unbalanced features of civs that are valid and perhaps could be addressed. Of course, personal preference also plays a role. However, none of what is being discussed here really qualifies as ‘bad design’ imo. What I would interpret as bad design is a civ that simply does not work properly. This means it either never wins or always wins. Looking at such helpful sites as aoe2stats, not a single civ has a winrate that would qualify as broken (either below 20% or above 80%, for example). If anything, most winrates straddle the 50% line, showing how marvelously well-designed AOE2 as a game really is, especially with so many civs. Not a single one can dominate everything, although on certain maps some civs work out better than others. Even civs with lower winrates are still not so ‘poorly’ designed that a good player can’t make them work.
TL;DR - Poor design =/= Unbalanced design. If all civs were equally balanced they’d basically be the same civ and that would be horrible. I don’t like certain bonuses either, but none are breaking the game or the civ too horribly.
Nevertheless, if I have to point out one civ I think works least, Cumans as a Civ make little sense to me. The only eco-bonus available to them is the two-TC play that you can use to get ahead your opponent, but with pressure or FC can be neutralized. Cumans don’t excel in early game if they go for a boom, but they also don’t excel in late-game due to missing bracer (making their ranged units, including Kipchak, rather underpowered). This means that Cumans using their eco-bonus seem to me to be only really able to push in Castle Age, before dropping off in imp - and clearly, looking at the winrate, most civs can handle what the Cumans throw at them. Perhaps Cumans should get bracer but the dps output of the kipchak slightly reduced.
People saying Goths are poorly designed, to me, are attempting to force Goths into the model of ‘well-balanced’ civs, which Goths simply aren’t. The Goths were never intended to be well-balanced but rather a civ that is just particularly good at one thing. The problem with balancing out the Goths is that this would make them too overpowered, because it would make it too easy to get to the spam. If you then reduce the effectiveness of the spam, what are you then left with? A spam civ that can get to a spam but it’s too weak to smash the opponent, like it does currently. The spam is indeed overpowered but it’s the task of the opponent to prevent the Goth player from getting there, and the latter is not an easy feat. In my opinion, that might be unbalanced, but not ‘poor’ design. If it were ‘poor’ design, the Civ would never win or always win.
Nor is Lithuanians broken. Yes, 4-relic paladins are OP, but they still die to enough halbs, and an opponent should also try to deny the Lithuanian player relics, like you would when playing against Aztecs. The 150-food bonus does give the Lithuanian player a lot of flexibility, but clearly this hasn’t translated into a 100% win-rate; so I don’t see it as particularly game-breaking.