I'm guessing this is prevalent in Japan not because "all the good domains are taken" but because (browser) support for unicode domain names (IDN) continues to be patchy, especially on mobile browsers which, as far as I can tell, are especially popular in Japan. I'm guessing they've long solved the problem of entering Japanese characters with a tiny keypad for things like text messaging, and I strongly suspect that entering names in their native script is much nicer than entering transliterated or foreign names.
Besides, punycode isn't especially efficient. I suspect you quickly run into the DNS hostname length limit as well, as the number of punycode characters required increases with the numerical value of the code points to encode. I'm too lazy right now to work out how many Japanese characters you can maximally encode in a domain name. :)
Note: I don't speak Japan, have never been to Japan and am generally out of touch with Japanese culture. My first language (German) only uses a small number of non-ASCII characters, so I'm not exactly an expert on IDNs.
Unicode domain names could be implemented with a one way hash function with ascii output. That would solve the length problem. Depending on the hash algorithm, collisions would be extremely rare, and i doubt many domain names are already registered that are equivalent to the output of most hashes.
I guess it would be similar in a way to tinyurl.com
This obviously would only be a good option if the hash didn't need to be reversible. And I don't really see why it would be, since the dns system itself isn't perfectly reversible.
Adding a second mechanism is just not going to work at all, I'm afraid. How would the application know which encoding you're trying to use? There's also no clean way to solve the reverse DNS issue with hashes.
Besides, it took long enough to get IDNs out "into the wild", and they're still struggling. IE6 still doesn't support them out-of-the-box, so there goes about half (or more) of your non-technical audience. I use exactly one site with an IDN regularly, http://öbb.at/ - the rail company here, and they actually advertise their other, non-IDN domain, oebb.at, presumably due to lack of browser support.
There was a discussion about this last year here on news.yc; that approach is not at all new, and "Totally out" is absolutely incorrect.
Search terms works especially in trains, simply because everybody has an internet-enabled phone that has quick keys to a search function. Saying that it's the reasonable next step from the current state of affairs, and assuming it resulted from the lack of good domain names, is quite a leap in reasoning. It's been a while since I payed any attention, but I never spotted a trend of startups using nonsensical/typo domain names like they did in the US.
Japan's internet culture is quite different from anywhere else, IMO.
Actually, a good portion of the "nonsensical/typo" domain names are international. While the US has its fair share of bizarre, cartoon-character inspired names, a healthy portion come from India, England, France, and others. Zoho (Indian, yet akin to a tasty Hostess snack cake), ShoZu (London), Plazes (Germany, so possibly a tad buzzed), and a good 30-40% of all Demo presenters. But, I digress...
I'd expect to see similar implementations coming back, on a larger scale. It's a lot of what John Battelle writes about as the next evolution of search. It's just with the SERPs flipped inside out (they find you) and plastered all over books, posters, etc.
QR codes work so much better, particularly for print such as the free commuter newspapers Murdoch gives to train travelers etc. The main problem is user acceptance/education outside of Japan & Korea. We did a small sample test (unscientific) with QR codes in Melbourne, AU: In the under 40s bracket, perhaps 1/15 could identify the QR code but only 1/25 could successfully use it.
I suppose if image recognition was good enough, there would be no need for QR codes? You could just photograph the url and the phone would act accordingly.
A disadvantage of QR codes might be that they don't work so well for human brains: if you want to establish your brand through an advertising campaign, a QR code alone won't help much.
The URL system is a dead end. There are just not enough good, or just reasonable, domain names out there. A drill-down method, for example, a simple search and a selection among a few alternative, looks like a potential alternative. I never used AOL keywords but this does look like a smart way to find a website.
There are tons of still good domains out there. Problem is, those filthy domain squatters (networksolutions, etc) have already registered them and demand $500+ for them! It is ridiculous that such practices are legal.
Arguing whether or not domain names in Japan are actually "totally out" feels short-sighted. Domain names are a scarce resource and will eventually hit a limit. The DNS system is fundamentally flawed in that respect, and I'm eager to see what will come next.
This article's proposal that search boxes are the next step sounds likely, but again, doesn't scale well, and has substantial search-engine risk. If Google drops your pagerank then not only is it harder to find you in a search engine, but all your marketing is useless because your call-to-action is broken.
In Japan, URLs are on the way out, while a U.S. company who's main asset is a bunch of domain names are buying a YCombinator start-up to further monetize them.
EDIT: I can't believe I just used "monetize" un-ironically in a sentence.
not a Japan exclusive: I saw a TV ad just last night on some local channel in San Francisco, that ended with a partial but branded Yahoo Search screen-shot, complete with the query pre-filled and a "Go" button, and a voice-over saying something along the lines of "the best to get in touch: search for [can't for the life of me remember the brand/product] on the web!"
Interesting post. My wife and kids have been doing this on their own, without any instruction, and despite my efforts to show them how to directly enter a URL in the address field. It just seems that it's easier for them to remember a search mnemonic than a URL.
I think Firefox 3.0 has added a feature to go in this direction. When you start typing in the address bar, it searches the titles and urls to match what you are typing instead of just the url. It's a lot easier to find pages that you visited this way.
In Firefox 2.0 the behavior that already exists seems to be like Google's "I'm feeling lucky" search. doda.jp is the first result from typing "doda" in to google.jp (but not google.com, interestingly) so probably if you have google.jp set as your default search engine in your browser then typing the brand name in to your browser should just work.
It's a very neat Bayesian DWIM thing.
Getting back to the OP, I think it's a non sequitur to go from "URLs have been downplayed in packaging and advertisements" to "URLs are out", though.
Besides, punycode isn't especially efficient. I suspect you quickly run into the DNS hostname length limit as well, as the number of punycode characters required increases with the numerical value of the code points to encode. I'm too lazy right now to work out how many Japanese characters you can maximally encode in a domain name. :)
Note: I don't speak Japan, have never been to Japan and am generally out of touch with Japanese culture. My first language (German) only uses a small number of non-ASCII characters, so I'm not exactly an expert on IDNs.