If someone's interested in a last straw, they probably planned to make real machines while still screwing people, like BFL.
They may have planned to keep dividends to themselves for as long as possible, while at the same time trying to dump their stock at good prices. Perhaps they didn't plan to make that many chips, they made just enough to keep a large chunk of IPO money and bring the stock price to the moon for a while. Too bad, they fired Howard Wang too early.
As long as they don't leave behind evidence that make them a blatant scam, they have some room to wiggle and maximize profit. After all, for what they've done they may just as well produce something real; they would really have sold more of their 3mil shares if they didn't at least believe they'd make real chips. It's a lot more profitable and low-risk than running a complete scam at this level of sophistication - it seems TheSeven did receive some legit questions on 65nm chips.
They have been extremely stubborn in taking advice on hardware design, so it is unlikely we would see real improvement in a matter of months.
In the current situation, buying in now means you trust 1) they won't run away with the money they have left 2) their current chips work. For the gamblers, things may get better in the short term if they finally become listed at Bitfunder or deployed just enough massively underclocked chips to get some real hashrate.