Well, LN Node Rankink is basically based on capacity and number of channels, which is a metric node operators look at anyway, so I don't think LN Node Ranking is adding to the problem if the problem exists.

Having said that I don't think there is a problem. There are multiple scenarios:

If you're a profit oriented node operator, you're interested in routing transactions. If you connect to only large nodes, you won't be routing much because they are already directly connected anyway. You need to connect parts of the network that are not yet connected. So you have an incentive to connect to small and medium nodes.

If you're not profit oriented node operator, you're interested to be connected well to the network. You could do that by connecting to large nodes, but large nodes usually have high fees because of the service they provide. You have an incentive to connect to medium size nodes and directly to the services you use or your friends you send and receive money from. This is to lower your fees to virtually zero.

So, in both situations you have a strong incentive to connect to medium and small nodes.

LN Node Ranking just simplifies how you think about the size of a node. Instead of looking at two metrics and comparing it to other nodes, you can just tell easily by looking at the 1 - 10 ranking.

It's a valid question. 

I've thought something on these lines. Incentives are incentives. If people are spending time and effort, risking capital, etc., I guess there's a natural ecosystem of marine life, whales to plankton. How you run a node could mitigate with more attractive fees / but less reliability than routes that charge insane fees.