Let's look at a hypothetical Widget industry. Some people make widgets out of materials they dig out of the earth. The digging process is expensive, as is the refining, but ultimately it yields such efficiency that the widgets are relatively cheap. Another group uses physics to split an atom, then harnesses the effects to produce widgets. This process is even more efficient than those that dig the widgets out of the ground and is utlimately even cheaper for the consumer.
Then you have this side of the industry that uses moonbeams. It's not especially efficient, but the government mandates that at least a portion of widgets must be purchased from the moonbeam users. The moonbeam widgets only work at night, because widgets are an on-demand sort of commodity. So, moonbeam widgets cost more and are only available for part of the time, so it wouldn't matter how efficient they are, without a way to save up moonbeam widgets for use in the day and when there's a new moon (i.e. no moonbeams).
So, this company comes along called Lunyndra. They have a new method of producing moonbeams. It's slightly more efficient than others in the moonbeam widget industry, but they still can't save up widgets for use in the day or during new moons, but they want to go into full production with this great new process.
The problem with their business model is pretty apparent from the outset. Industries are forced to buy moonbeam widgets, but only a little bit and only during the night and only if there isn't a new moon. Without a means of storing up the widgets for use in the daytime, the current supply on the market for mandatorily purchased over-priced widgets is saturated. Lunyndra's production process is expensive and producers wouldn't really need to buy their more efficient units because what is currently being produced is sufficient to meet the artificial demands as established by the Government.
For these reasons Lunyndra can't secure financing in the private sector. Banks won't lend to them because they don't have a sound business model without a component that allows moonbeam widgets to be stored and used during the day. Venture capital is staying far far away for the same reasons.
In steps the federal government. They guaranty a private loan and suddenly banks are more than willing to lend the money...after all, it's a 100% riskless loan now. The President calls it a "good bet" and was actually considering upping the loan from $500+ million to closer to $1 billion.
You already know that this hypothetical industry is the power industry and moonbeam widgets are solar energy and Lunyndra is in fact, Solyndra. You should also be aware that the limitations effecting the Moonbeam widget industry also effect the solar energy industry (in reverse of course, with energy being produced in the day and not at night or when there are clouds, etc).
The attitude of the President and his administration is actually remarkably normal for people not familiar with the funding and operation of business ventures. While there is risk involved in any business venture, smart decisions are vetted and analyzed quite a bit more than a roll of the dice at a craps table. Before a bank loans money on such a scale, before venture capital becomes available, a road to profitability must be shown. Nothing of the sort was done here. Instead a politically connected donor wanted free money and got it based on someones "gut" or the fallacy that business is like gambling.
Why shouldn't government be involved in business? Because Government is for sale and their asking price is a lot cheaper than the costs to you and me.