EDIT: Ragnorock pointed out one mistake which I am actually quite happy to acknowledge because it basically makes it both unreasonable for us to expect a discount if he pushed the next price break and unreasonable for him to really take advantage of anything without a fair associated risk. Because substantial edits were made, I will preserve the old post in code tags at the bottom (to make it shorter) for historical reasons. I'll happily take the blame for my mistake.
I also would like to say that this is/was not suppose to be threadcrapping. I have in fact always been in favor of the OP and trying to help resolve any potential issues, though I admit that I am bringing up potential issues that would otherwise probably not have been considered. Still I think it's a good thing because it's an element of group buys that hasn't been considered too much in the past from what I can gather and it's good to try to address it rather than have more of the controversy that was had in his past group buy.
The issue for those who haven't seen it yet is with SP's sweet spots which don't make sense. If a sweet spot is reached, the person in charge of the group buy can get free keys up to the next price break but can't offer a discount for that price break or the keys aren't really free. This put him in an odd spot where he has extras he can try to profit with but without being able to give people the discount without taking the loss himself.
---
Because of where the sweet spot is between 100 and 250, you'd be taking up only 29% of the orders (57/196 as shown below), not (250-139)/250 = 44% which is the nearly half you mentioned. In the end this really won't make a big difference in any case though, it's just to address that statement.
The amount needed to make the 250 price break is 196 keys assuming no price break at 150 (250*2.96/3.78 = 196). If there is a price break at 150 this discussion is moot since we'd just aim for 150 keys, and we probably are already at the sweet spot for it. We really should ask about this when ordering.
We have 139 orders already.
196 - 139 = 57 additional orders needed
So you would have to pay
57 * $3.78 = -$215.46 (this is the one mistake that changed everything; he does not pay the price break amount but does get the extra keys)
This would give 250 - 196 = 54 free keys
and 57 paid keys which could be sold for markup
54 + 57 = 111 total for resale
If you sold them for $5 each it would be
111 * $5 = $555
Then subtract what you had to pay for the 57 paid keys
$555 - $215.46 = $339.54 potential profit without sharing the discount (emphasis on potential, i.e. not guaranteed)
The opportunity cost is -$215.46 if you manage to sell none.
You'd have to sell 43 (not done easily) of them to break even and you'd have 68 left to try to profit with. Not something I would bank on as I don't see there being a market for so many nyancat backspaces in our community. If you did manage to sell them all somehow, it's not a bad return though. Keep in mind that if he sold them for less he would need to sell that much more and the real limiting factor is the size of the market.
If discount is shared
139 * ($3.78 - 2.96) = -$113.98 that would have to be shared
$339.54 - $113.98 = $225.56 potential profit without sharing the discount
The opportunity cost is -$329.44 if you manage to sell none: -$215.46 - $113.98 = -$329.44
You'd need to sell 66 of the 111 to break even and you'd have 45 left. This is extremely risky and not worth doing.
---
The latter case, where the discount is shared is available to anyone, since anyone could buy up, it's just the first that is exclusive to him. In both cases, it's possible for him to profit but it's not an "easy" profit so if he wants to try I think it's more than fair.
The actual best thing would be if there is a 150 price break, and I imagine there must be one since there are price breaks at 150 for other keycaps, unless it's different for dye sublimation.
Foreword: Before reading this, know that I am not angry, I just want to go through the numbers for transparency.
Because of where the sweet spot is between 100 and 250, you'd be taking up only 29% of the orders (57/196 as shown below), not (250-139)/250 = 44% which is the nearly half you mentioned.
The amount needed to make the 250 price break is 196 keys assuming no price break at 150 (250*2.96/3.78 = 196). If there is a price break at 150 this discussion is moot since we'd just aim for 150 keys, and we probably are already at the sweet spot for it. We really should ask about this when ordering.
We have 139 orders already.
196 - 139 = 57 additional orders needed
The price at 250 keys is $2.96, so you would have to pay
57 * $2.96 = $168.72
This would give 250 - 196 = 54 free keys
and 57 paid keys which could be sold for markup
54 + 57 = 111 total for resale
If you sold them for $5 each it would be
111 * $5 = $555
Then subtract what you had to pay for the 57 paid keys
$555 - $168.72 = $386.28 potential profit with sharing the discount (emphasis on potential, i.e. not guaranteed)
The opportunity cost is -$168.72 if you manage to sell none.
Pretty high opportunity cost (this is bad) but with reasonable return. You'd have to sell 34 (not done easily) of them to break even and you'd have 77 left.
If discount is not shared
139 * ($3.78 - 2.96) = $113.98 extra from keeping discount
$386.28 + $113.98 = $500.26 potential profit without sharing the discount
The opportunity cost is -$54.74 if you manage to sell none: $113.98 - $168.72 = -$54.74
Very low opportunity cost and very high potential return. You'd only need to sell 11 of the 111 to break even and you'd have 100 left.
---
In general, if you do not share the refund it puts you in a very awesome position.
If you do share the refund it's hard to say. If none of the extras sell you are down -$168.72 but you have the potential to make quite a bit if they do sell but I wouldn't bank on it, at least not in the short term. It would really be playing the risk/reward game. You could lose a lot but you stand to make a lot if you can find buyers, but this is "normal" when trying to do business right? If you want to make profit you have to take risk.
So what's the issue? If you don't want to order up to the next price break that's fine. We get no discount and you take no risk. If you do order up to the next price break and don't share the discount, we get nothing and you get the chance to make a high profit with minimal risk because it would be cushioned by the unshared discount.
This is both good and bad though... on the one hand, since the other people in the group buy are 71% responsible for meeting the 196 orders (100% - 29%), it sucks that we don't see a discount. On the other hand, if you buy up and we don't get the discount, it doesn't really hurt us, and we can be nice and give you a chance to make money... why not give a geekhacker this chance, right?
If you do as dorkvader said and share some of the discount it might reach a nice medium where the buyers get some price reduction (for the opportunity provided and as deserved for being in the group buy), you have modest opportunity cost and modest potential return. In the case of sharing half the discount the opportunity cost is about -$100 and the profit is around $450. Not all would sell; 20 sales are needed to break even with 91 left over to try to profit with.
The problem is that in every situation, if Ragnorock buys up, he starts by going into the negatives, but this is the risk associated with potential future profit. The question, to which there is no real answer, is what amount of risk and profit is fair? And thus the whole problem and its associated drama. If we fight too hard for a discount he just won't buy up and we get nothing. If we don't fight at all we don't get the discount we deserve, but, at the cost of like $0.82 per person (times your number of backspaces), we give him the chance to make a nice return.
The further I get into this, the more I think that he should just do it and not give us the discount really, and that's coming from the guy with the most backspace orders. This is analogous to the "everybody in the world gives somebody a penny and he's rich and they don't really lose out on anything" scenario.
The actual best thing would be if there is a 150 price break, and I imagine there must be one since there are price breaks at 150 for other keycaps, unless it's different for dye sublimation.