Royalties are simply the most rational way of paying songwriters - the ratio of successful songs to unsuccessful songs is so large that it makes no economic sense to do it any other way, and the randomness of the music market prevents any real sort of cost/benefit analysis - even huge stars can flop, one-hit wonders can pay off spectacularly, and you have to take hundreds of risks before you find a profitable performer. A successful artist has to write a large number of unsuccessful songs before they can profit from one, regardless of the relative merit of their work.
How would you feel if you wrote dozens of good programs, but only got compensated for one if the client hit a improbably large sales target, and you only got a small lump sum anyway? The music industry is a bitch, but it works like it does for a reason, and its crimes are crimes of disproportion rather than inherent evil.
It is one thing to rail against abusive organizations, but you have to think about the little guy here. Music as a good is worth what people will pay for it, nothing more and nothing less, it has no inherent value and it is impossible to know in advance just how much it is worth. You have to allow the possibility of profit, however improbable, or the quality of music will decrease even further beyond your lofty standards. Royalties are a merit-driven method of compensation, or as close as we can get allowing for market distortions from publicity campaigns, and are still very relevant (more than ever, in fact) in the post-CD age. Royalties are the only method of compensation that make sense with a lack of scarcity - either through subscription or advertising, royalties can be paid from online radio, with (at least the possibility of) accurate and fair accounting since it is easy to track exactly how many times a song has been heard.
And yes, of course people used to get paid for specifically commissioned work, BEFORE THERE WAS RADIO. Royalties have been around as long as modern music has. A patronage system simply doesn't make sense anymore because of the broadness and tastes of the market - you make way more money selling to crowds of teenyboppers or hipsters than to a single customer, and you need the sieve of the market to find talent amongst the millions of musicians producing today. Plenty of terrible work has been commissioned over the centuries as well - think of how many classical composers, or artists of any kind, that you can name from the past 500 or so years, compared to the number of artists you can name from your lifetime. Aside from being completely apples to oranges, there is a serious survivorship bias in comparing the cream of the premodern crop.
It is impossible to determine the true profit of a song in advance. Engineers can get a lump sum up front because it is much easier to determine the objective worth of their work; with songwriting, success is mostly random. Some professional songwriters can hit on a formula that might be successful within the context of a given trend, especially with the backing of media corporations, but fashion changes quickly and there are very few who are able to be consistently successful over the long run. A piece of art is priceless, or valueless, in both senses of the word - it is simply as valuable as people are willing to pay. If you write a program or build a structure, you can estimate to a pretty fair degree how much value you'll be able to extract from it, but there is such great variance with music, even with millions in marketing backing a project, that there is no way to pay people up front. I suppose there are songwriting groups and in-house guys that can make a salary, but outside of highly commercial productions that's just not how people write music.
If current practices are unsustainable, then popular music as we know it may be unsustainable. It is hard enough to make a living as a musician now, lots of mildly successful musicians still have day jobs, but without the small amount of revenue from royalties many might just give up. Also, being a songwriter doesn't necessarily mean being a performer, or a touring artist, so you can't just say 'give your music away for free and make it up in t-shirt and ticket sales.' Sure, there are site-specific jobs like in-house composer for films and games or commercial jingles, but they are the people least likely to receive royalties anyway, having already signed away rights to their employer. You could sell rights to use your songs in advertisements, but you need a certain degree of exposure in the first place to have enough street cred to be worth associating with a product, so that money tends to go to those already successful.
Note that this isn't an issue of selling CDs - in fact, royalties from streaming sites might just be the primary source of income in the near future with the popularity of places like last.fm and Spotify. Proper allocation and collection of royalties is more important than ever these days. Don't throw the baby out with the bathwater because of these jerks, we're just getting to the point that the granularity of royalties makes sense, now that actual accurate playcounts can be generated (and individual songs purchased rather than albums), resulting in fairer distribution of collected fees based on harder data than chart sales and radio playlists.
Remember how Time's 2006 Person of the Year was YOU? And everyone hated it and thought it was a terrible choice, because user-generated content is often idiotic, base, lowest common denominator, whatever (not trying to be biased, that's just how I remember this place reacting)? That no one cares about your stupid band, your podcast, your profile, or your feed? That the average web user's narcissism might not be the best method of content generation, that social networking concepts were being shoehorned into places for sake of bandwagon jumping, at the expense of added noise and reduced quality content?
In other words: do you really need to put your face on Mario's body? Does that truly enhance your game playing experience? Should game storylines be written around a shallow method of providing a surface-level customization for added 'personalization'? The article takes issue with inappropriate uses of character customization, a trend that has begun to spread from its traditional place in choice-and-consequence RPGs (Fallout, not Final Fantasy) to pretty much any kind of game (often, seemingly randomly), a move that has begun to change the manner of storytelling in video games. The author thinks that this customization, in attempting to bring players further into the narrative, is actually alienating them by presenting them with meaningless choices, confusing identification with understandability, distancing the player from whatever intent the storyline has by introducing surface-level similarity at the cost of a more coherent characterization of the game's hero. Think of the Time cover writ large; a mirror over the face of a video game's protagonist. If there were a technology to easily switch your face with that of an actor's in a movie, would that help you understand the film or extract any additional meaning from it? Does every story need to be turned into a choose-your-own-adventure with branching paths, at the cost of a greater unifying vision? And what purpose does customization serve in the cases where there are no branching paths, when it is thrown in because of market trends?
Assuming you care about video game narratives at all, let alone as any sort of art, of course. I don't think slashdotters are the right crowd for this kind of article. You need a few years of undergraduate literature and film classes to write like this, I don't think this place has the background in narrative theory necessary to be interested in the points this guy is making. Frankly, you guys should consider yourselves lucky not to understand this guy's writing, it probably means you are gainfully employed.
"Pok pok pok, P'kok!" -- Superchicken