because
As a knowledgeable programmer, your internet experience with DNS and browsers is very different from others, because you know what you're doing. Many people do not.
Should we teach them? Or should we try to control them by controlling "default behavior"?
Many people reading HN are probably too young to remember the browser wars and the battle over the "default browser" setting. This is why IE triumphed over Mozilla. And as any web developer knows, the web has suffered for it. Users who knew what they were doing could change the default settings. But that did not make the difference. Chrome is highly configurable. But that will not make the difference. Default settings are what matter the most.
The success of ICANN relies on a default setting. It's called "root.hints" or something similar. If a person running a DNS server (that could be anyone, including the end user) changes this setting, they can bypass ICANN and these landgrab gtlds.
In the distant past, "alternate roots", an early reaction ICANN's abuse of power, failed because they were trying to do what ICANN is now doing: make money by selling registrations in additional, redundant gtlds. Perhaps the "alternate roots" of the future will be noncommercial ones that subtract gtld's, not add them. For example, maybe we as smart users will choose to use a "pre-landgrab" root that does not have all these silly redundant gltd's that try to capture whatever unorthodox string a novice user might type in a browser.
Domain names are important, but always remember that those who seek to control the browser can easily subvert DNS. Given this decision from ICANN to sell out (which others, not just Google and Amazon, are following: e.g., the founder of BlackHat is running the newgtld program for ICANN; the author of BIND is partnering with a domainer to run a registry), we're going to see some aggressive moves from commercial authors of browsers to control type-in navigation. Of that you can be sure.
Again, the question is whether we want to teach users how things work, or whether we want to let them remain ignorant so we can control them. In the big picture, ignorance does not help the web. Though it may benefit certain vendors.
There's money to be made by owning a domain like amazon.net. You would be amazed at how many people type searches and erroneous strings (e.g., .net when they mean .com) into the Address Bar.
Just because we can't imagine anyone doing this does not mean they don't. They do. That's why numerous domainers are millionaires, some of them a hundred times over, and that's why they are salivating at the thought of running a registry. Zappo's had shoes.com. Now it belongs to Amazon. Amazon knows the power of domain names. Users do type in the Address Bar. We might not know who these users are, but they exist. They have made a number of domainers very wealthy. And Amazon has paid millions to acquire domain names, no matter how ridiculous we may think domain names are, e.g., they paid a premium to acquire a.co
But you're wrong. When people go to buy a computer, most know what RAM is, i.e., they've heard of it. It's because advertisements always mention it.
You could argue ICANN extorts money here by putting pressure on companies to apply for gtld's in order to protect their marks. Many companies do not want to apply. For example, the world's most valuable trademark is missing from this applicant pool: Coke. There's no reason ICANN has to create gtld's that match trademarks by putting them in the ICANN root zone. It is domainers that want new gtld's the most. (For the simple reason why, see below.) How hard is it to do 2000 trademark searches? Essentially ICANN is profiting (they get a small fee for every domain name registration that is later sold) by selling the rights to use someone else's trademark.
Then there is the extortion that is carried out by the registries and registrars and domainers. Companies are forced to purchase and renew registrations for direct matches and typos of their trademarks in many different gtlds. More tld's equals more money. Also companies register and renew names in the form [companyname]sucks.tld. That is why you see some applications for a .sucks gtld. These defensive registrations are the bread and butter of the domain name "business". Again, they have no right to sell use of these trademarks. But they do. And they make a handsome profit from the "protection money" that must be spent by companies as a result.
That said, you do have a point: If you are a large company, $2000 and the fees to register domain names in lots of tld's, or to file UDRP's, are essentially nothing. It's easier for companies to just pay the fees rather than try to figure out how DNS actually works and how ICANN and their followers manage to pull off this scam.