I feel that completely practical question are on-topic for SO, research-level question about algorithms and data structures are for TCS.SE, and non-research-level questions about algorithms and data structures are for Math.SE. I personally don't understand how the two top voted on-topic questions: "Can you explain the difference between NP-Hard and NP-Complete?" or "Why can't a comparison-based sort be less than O(n log n)?" can be considered a practical algorithmic/data structure question. I think the supporters of the proposal should give examples of on-topic questions which will distinguish the proposal from already existing related sites by being off-topic on all of them. I don't think it is a good idea to start a new site if it will not cover a reasonable amount of questions which are not covered on already existing sites.
We see a some amount of cross-posting between MO, TCS, Math.SE, SO. (I don't not know the official view of SE people on having multiple copies of question on different sites when a user is trying to maximize the chance of getting an answer.)
Having too many sites will divide users and question, it will drain users from already existing sites and there won't be serious communities. I feel that people follow and commit to the proposals when they are interested in the proposed topic and without considering whether having a new site for the topic is a good idea.
ps: I think in long term the SE people need to find a better way of reducing the duplicate copies on different sites and increasing the coordination between them.
Here is an idea: we can think of each site as a view and questions on each site as a super-tag combining the copies of the question and the answers from different sites in one place and letting users see only the answers having specific super-tags. Deleting a question from a site will remove the corresponding super-tag. I think in that case having largely overlapping sites should not be a problem. I am sure there are lots of things that SE people have to consider and the this is not a through thought idea.