This specific episode was somehow unique in the sense that this time the plans really were changed remarkably in key points multiple times as the project went on, and the host of the show looked really doubtful of the chances of the owners ever getting a decent house. As usual in the series, the end result indeed was a decent house, unique for sure but not in a weird way, and definitely not ugly to look at (I wonder if they just leave out all the miserable failures or if the house builders just are so darn lucky).
Software or system development projects have been compared to all kinds of more traditional projects in the history, and software industry also has tried to take influence from other industries (like lean that originates in car industry). The inherent differences between software and its comparison points that might make the metaphor lame are debatable, as cars and houses (and maybe even more so the huge projects like bridges, dams etc, not to mention nuclear power plants) really are quite complex and technical these days.
The text book recipe for success is to have a rigorous working process that is not abandoned when difficulties arise, to avoid touching legacy parts when possible and to have professionals that are knowledgeable on the task they've taken on all levels of the project organisation. Many times it just doesn't go like that, I think. Many times there will be a sort of a cowboy mentality, "agile" or perhaps better called anti-agile (as it does not have much anything to do with the definition the signees of Agile Manifest had in mind) way of working. Many times it is indeed the legacy parts that need to be touched, which is harder, more expensive and more error-prone. Many times the key persons might be professionals of some other trades than software (no-one really expects non-coders to code, but surely any person with a technical mindset can be an architect, and financial skills are enough to lead the project because then the project surely will be on budget, right?)
The success rate of software projects is not very flattering. I've not got any recent figures at hand, but Robert Kraut and Lynn Streeter referred in 1995 in their work "Coordination in Software Development" to Robert Blazer's statement dating as far back as 1975 (!) and said that software still is unreliable, delivered late, does not respond to changes, doesn't hit the performance target and is expensive. Well, has that changed since then? I cited that paper in my MSc thesis in 2007 and my conclusion is still the same: that does still happen in large numbers.
Looking at the news here, I'd say that the success rate for building projects is not very good, either, especially the ones that are not large and demanding enough to clearly require the best practices and the best people. Just like the resulting software is likely to have security issues, the resulting buildings (at least in this country) are likely to have issues with moisture and/or unhealthy athmosphere inside the building. Neither of those would be hard to avoid, though, either by thinking security already in the design phase (for software) or applying basic building physics in the design phase and carefully sticking to the plan through the building phase (in construction work). But like in the TV series, when asked about the expected end time, it can be found out that there is no detailed schedule, and the project owners have the habit of changing the design as they go along (each change perhaps leading to other changes), and the project lead might be technical, but not exactly from the right branch of science...
As for the couple that featured in the TV show, I have absolutely nothing to say to how they dealt with the house they were building for themselves - it is solely in each one's own decision and risk. After all, they were amateurs when they started the project (as much as I am what comes to building a complete house), and should thus be allowed to learn as they go. However, the same mentality and way of working would be very strongly objectible if it occurs in the context of professional work.
So, what all this babble boils down to, I assume, is that some software projects are either led or completely run by a bunch of amateurs, or a bunch of people acting like amateurs, which is sad and unfortunate (both for the industry itself and the users of the software). I'm not into poking fingers at people, and I must confess that I've done plenty of amateurish mistakes so far and likely will discover new amateurishness in my actions in the future. Live and learn, or should I say fail and learn...
Since my sources are a bit dated already, I did some quick googling on the current situation:
Why Projects Fail - Facts and Figures
The Failed Record of the Software Industry
It would seem that there, indeed, has been some improvement, and the agile movement really seems to be able to deliver what they've promised. However, the papers referred in the above sources still contain the same complaint and worry: that it is the people who make the projects fail. People, who do not act in a professional way.
From a certain angle computers creating their own software sounds like a good idea, even if it is quite sci-fi sort of an idea...
Post addendum
Some more survey results:
http://thisiswhatgoodlookslike.com/2012/06/10/gartner-survey-shows-why-projects-fail/
http://www.galorath.com/wp/software-project-failure-costs-billions-better-estimation-planning-can-help.php
And then yet later on (13th Nov 2013):
A good example of a gigantic failure in a gigantic project is the US healthcare.gov project that has been under scrutiny lately.
