Torrents also increase in speed as the number of peers increases. swarm Is the same? This is because dividing the data into chunks is the same.
In swarm there are at least two factors to consider in this case. Rather simplified:
In a torrent, a "saturated" peer (seeder) would mean a peer that has all the chunks of that particular torrent.
In swarm, nodes are responsible of certain subsets of chunks of all the data chunks that exist in the whole network. A "saturated" peer would here rather mean a peer that has all the chunks it's responsible.
In the simplest mode, swarm nodes takes an opportunistic approach to storing data; they store the data that is most profitable to them. This is the content they currently get the most requests for, because their earnings come from the number of chunks they deliver.
Because of this, nodes will tend to keep chunks which are also outside their responsibility, if they notice those chunks are in high demand.
That means that popular chunks will tend to propagate more, which ultimately means that any path to those chunks will tend to be shorter. And shorter path means more speed.
That said, there are other factors that might influence the speed of delivery, of course, like geographical distribution and node bandwidths. There considerations aren't currently part of the consideration in the swarm architecture. These may just as well be a factor with torrents, but with torrents, since you connect directly to all peers that have the particular chunks you know you want, it's trivial to detect which ones who suck.