I was doing the math on this reversal idea earlier, and I think there is an aspect being overlooked.

There were five tranches of shares issued, starting at 5k and eventually reaching 40k.  Just for demonstration, I'll focus on shares in the first tranche.

If I understand correctly, the contracts were sold for 1 BTC.  This generated approximately $5.12 for Giga to spend in the normal economy on hardware and other business costs for the operation.

Counting up until the last dividend that was actually paid on GLBSE, the share generated a total of about 0.48 BTC for the owner, leaving 0.52 BTC to pay to the owner in order to negate the original contract, according to what has been proposed here.

But due to the dramatic increase in BTC value, Giga would actually have to spend about $7.01 this week just to cover that 0.52 BTC, significantly more than the dollar capital that the whole share generated in the first place to actually buy equipment, space, and electricity!

One other perspective that you might or might not find to be relevant: it is easy to calculate the fair market value in dollars of each dividend at the time it was paid.  The cumulative FMV of all the dividends that were actually paid through GLBSE is $3.68 on that trache of shares, or 72% of the $5.12 FMV that went in. If you sum all the way through the Dec 10 dividend payment in the queue, the total rises to about $4.84 in dividends, or 95%.  

That's assuming that each dividend is immediately cashed out into dollars.  If a holder of a share in that tranche has kept the whole 0.48 BTC in dividends in their wallet, its FMV now is $6.49, still a return significantly over 100% of the dollars it took to acquire the share at the time. Naturally, that ignores opportunity cost, etc, but it's till something interesting to know.