[Proofrock]

So let me distill what I can from the responses and my thoughts regarding the argument so far.

(1) There is some question about the assumption that MTT seating is random (i.e., that it's possible to sit at the same table with yourself at any point in the tournament). There is anecdotal evidence that seating is entirely random at the beginning / with table changes, but nothing hard yet. This, obviously, would have an effect on the validity of the argument (or, rather, drastically alter the frequency that one ends up at one's own table).

(2) OP's calculations jumped from 2 accounts to 3 accounts for the later large field argument -- this increases the number of paired combinations from 1 to 3, thus magnifying the effect by three. Justification for this?

(3) OP's calculation only considered starting arrangements (by design, it seems). Since better players tend to go deeper than weaker players, and there's an unstated assumption that better players are more likely to adopt such measures (correct?), we expect the likelihood of sitting at a table with yourself to increase as the field decreases. If this is the case, then the effects of (2) and (3) may partially cancel each other, yielding an approximately accurate result for somebody using just two accounts (instead of 3, as calculated).

(4) Something that I haven't seen discussed -- is the advantage of sitting with yourself greater early in the tournament, near the bubble, or later in the tournament? i'll with-hold my argument for the relative advantage of each, but the answer to this question may help indicate the degree to which dual-playing would have the most meaningful negative impact on the EV of solo-players. I can elaborate if this is unclear.