Except that's exactly what's not happening. Take this case. Suppose that they now start forcing N% of contracts to IT businesses run by women. Now there aren't many such businesses (regardless of the reason), which means that competition for that N% is going to be lackluster - heck, it's spelled out in TFA, pretty much.
You don't know that -- all you've done is take a snapshot of a single instance in time, saw what you perceive to be a minima, and decided that is the nature of things now and forever.
Conceptually, affirmative action is like a social algorithms that endures local minima, but with the design that an eventual equilibrium (or global maxima) will be generated. We have a lot of algorithms like this in computer science[0]; many optimization strategies are known to generate poor intermediary results, with the end-product being either the correct solution, or the best solution derived from the algorithm execution (ideally within some known, expected bounds)[1]. Genetic algorithms work this way: a single generation, viewed on its own, may be at a local minima, and thus an extremely poor solution. However, viewed at the end of the run after multiple generations, a better solution can be obtained (either a high local maxima, or a global maxima -- if one even exists).
Or, put in different terms, look at one stock market index. There are many local minima over the course of the index, but the overall picture is one of growth.
Capitalistic theory argues against you, using your very points. N% may be lackluster for a specific, given contract[2]. If the contracts are sufficiently lucrative, and there are financial benefits to be obatined from that market, then more organizations will desire to enter that market. In this case, the "market" is the artificial construct of "IT businesses owned by women"[3]. If there is money to be made, then more entrants will fill the market. If you knew there was a field where you had a significant bidding advantage because you (for sake of example) had green eyes, would you not consider entering that field and reaping the benefits?
As more female-owned IT businesses bid for such contracts, competitive pressures will start to take effect, to the point where the local minima of "N% lackluster" is nullified. Thus, the concept is:
- 0. Endure potential local minima, while aiming for a global maxima,
- 1. As more entrants try to take advantage of the market, allow competitive pressure to improve them towards a quality maxima, and
- 2. Achieve your original end of social engineering the market to increase the social structure(s) needing improvement (in this case female owned businesses).
You can't observe a local minima and then decry the entire algorithm. When it comes to social engineering, the algorithms often require multiple generations to achieve their ends; there is a good probability that it won't happen within a single lifetime. To use a local minima to judge an algorithm is illogical and narrow-minded. In this specific case, I see it as laudable to attempt to further engage women in business ownership -- there is a glass ceiling, and simply allowing things to progress as they have for the past 70 years won't change anything.
So let the algorithm run. The competition right now may be lackluster, but as more potential female business owners learn of the opportunity, they'll enter the market to get a slice of the pie. As soon as there are multiple female owners in the market, competitive pressure will be to out-bid each other, improving the process to the point where they are indistinguishable from their male-owned competitors. They still have to meet all of the requirements of the tender, and are still expected to produce results -- and over the long term, as competitive pressure and increased female ownership takes place, an equilibrium will be achieved whereby such actions won't be required anymore, as female business owners will be fairly commonplace[4].
---
[0] - I've been rather disappointed in the Comp.Sci. literacy of the average slashdotter over the last year or so. It feels like the MCSE types and scripters have taken over, and the kernel hackers and algorithms people have left. Case in point: the article the other week on the new Blackberry OS. Instead of a rousing discussion on the real-time nature of the QNX Neutrino kernel, the discussion was entirely on RIM's business model. Sad.
[1] - My M.Sc. thesis was precisely on developing such an algorithm to minimize the cost of data synchronization/cache coherency, where cost was measured by the inverse properties of a) the cost of data transfer within the network itself, and b) the cost of using out-of-date information (i.e.: you can reduce network costs by synchronizing less frequently, but then the chances of getting stale data increases; inversely, you can reduce the cost of using stale data by synchronizing more frequently, with a rise in cost of network data transfer. An ideal solution finds a minimal equilibrium, and my research centered around a possible algorithm to reach such a minimal state over time). As such, I'm intimately familiar with such algorithms.
[2] - You've made a huge assumption here that female owned businesses would produce inferiour results. Citation needed.
[3] - Note that the article didn't say that the women were the IT workers for the project -- simply that the companies involved were owned by women. This goes to [2] above -- you have no evidence that a winning female-owned company would perform any worse than a male-owned company; such a female-owned company could very easily employ nothing but neck-bearded UNIX geeks of the same calibre you presume (without evidence or citation) a male-owned IT company would have.
[4] - With the side-effect being to ideally remove the glass ceiling. If 50% of privately-held companies are owned and run by women, then any individual woman will have less sexism to worry about, as (ideally) women aren't going to automatically discriminate against women.
---
Yaz