Hacker Newsnew | past | comments | ask | show | jobs | submitlogin
Amazon's short URL: http://a.co (a.co)
64 points by shawndumas on May 24, 2013 | hide | past | favorite | 61 comments


One of the main reasons I used short URLs before was twitter, but they shorten* all of them for you now.

*Unless the original URL is less than 22 characters, in which case they conveniently _lengthen_ it for you. Which includes many of the unshortened links I post. I even recently set up my blog so posts have a ~9 character permalink. So twitter would award me with about a dozen bonus characters.


I especially hate that behavior as occasionally I'm talking about a domain name (maybe one that is having a DNS issue), not a URL, and Teitter insists on making it a lengthened URL. I've seriously taken to putting zero-width space characters before the .com to break its detection.


What do people use short URLs for nowadays? I used to run one that was "dynamnic" in the sense that you had a number of services that it could resolve and specific data you would send the services. But it never gained traction.

/g/search+query /bing/search+query /amazon/product+search /g.images/image+search etc.


I use them when sharing links to things that have seriously long URLs. For example, a link to Google Maps looks like this: https://www.google.com.ar/maps/preview#!q=Great+Pyramid+of+G... (291 characters long).

And other people use services like goo.gl to get analytics on links.


I mean, you mentioned Google, so yes, expect 200-300+ character links for the most simple of urls, with about fourteen layers of different metrics baked in.

Nothing worse than trying to get a link from Google search results.


I guess that's what I'm asking. The only thing I know it's really useful for is making small QR codes. Of course, a lot of people would disagree about the utility: http://picturesofpeoplescanningqrcodes.tumblr.com/


A coffee shop in my city uses Stampt[0], so I see someone scanning a qr code nearly every day. Still, that tumblr is pretty dang accurate.

[0]: http://mobivity.com/stampt/


I use them when sending URLs through text messages.


one major use-case is tracking. The short-url goes to a redirector where the visit gets tracked. That's the most annoying part of short URLs.


I use http://shoutkey.com/ all the time for conveniently moving URLs between computers, or communicating them to others.


Short URLs are still useful for sharing links in email (to avoid email encoding or line wrap problems) or verbally to other humans.


Twitter explains their use of t.co on the root of the domain:

http://t.co/


Their explanation is a lie. They do it for tracking purposes.


as a quality signal for surfacing relevant, interesting Tweets

Doesn't that mean tracking?


And the owner of http://a.pl still hopes for a call from Cupertino.


Didn't they get the call last year: http://www.pcmag.com/article2/0,2817,2409669,00.asp


that story makes me sad...


I might have purchased ap.pl or aa.pl instead. (Which one is their stock identity?)


AAPL


As expected, Google own http://g.co and twitter own http://t.co yet http://f.co is available.


Why would they need/want f.co? They already own fb.me. f.co is one letter shorter and isn't as easily identifiable as fb.me.


The same reason Google owns Gtaxes. It might be useful in the future and better to have it yourself than someone else? I'd imagine FB would want to squat on it at least.


f.me would be excellent.


Unfortunately .yu ccTLD expired.


Look dk domain http://dk resolves even without the subdomain.How's that possible?


Nothing special about a tld resolving. You can put an A record on anything. Many companies set up internal DNS names like "go" or "wiki" so that they resolve to helpful internal tools. This is like a public version of that.


The global top-level domain is a dot (.)

Point your browser to www.google.com. and it will resolve the same.

com. and net. and dk. are top-level domains, but still are "subdomains" under the global ".".

The administrator of the dk tld just pointed a dns record at it.


Just wonder whenever http://./ could exist.

Theoretically, I suspect, this should be possible, as an empty-named A/AAAA records on root DNS servers, although dig(1) and host(1) failed to produce such query for me.


It's owned by the danish top level register. I guess they just added a dns entry for that.


It is "just a domain". Top level at that. If you control it, you can point it to an IP.

There are more, http://shii.org/tech/tld.html


The "to" TLD used to resolve to an URL shortener, producing http://to/xxx urls.

Now it still resolves, however the server doesn't accepts http connections anymore.


The browser is automatically putting "www." in front of "dk" allowing it to resolve. Although the root DNS zone (the implied dot at the end of DNS names) can have A records in it, the current policy only allows A records for (some) nameservers. As a result, the new gTLD program, for example, does not currently allow "bareword domains" because you can't put an A record for the domain by itself in the root zone. (The policy doesn't allow an MX record, either, so email to someone@shop wouldn't be possible without a policy change.) The policy is aimed at keeping the root zone simple, and minimizing the volume of changes.


I don't know a lot about DNS, so correct me if I'm wrong, but this looks like a case where the TLD does have an A record:

    dk.       86400	IN	A	193.163.102.24
In fact, www.dk resolves to a different IP:

    www.dk.   86400	IN	A	212.242.42.44


You're right -- dk. has an A record. It's coming from the dk name servers, though, not from the root zone. (I should have run the query the first time.) You're also absolutely correct that dk and www.dk resolve to different IP addresses. So the root name servers are returning a redirect to the dk name servers, which are providing an A record. (You can see it with "dig +trace dk.") The dk name servers can return whatever they want. I believe the behavior is undefined, though -- where a zone and a parent zone have records in common, they should be consistent. So you may get different behavior from different resolver implementations. (This may help to explain why some people see something at http://dk and some don't.)


Doesn't resolve for me.


This is strange. I was able to resolve the IP using the host command. But curl and ping won't work.

    $ host dk
    dk has address 193.163.102.24
    dk has IPv6 address 2a01:630:0:40:b1a:b1a:2011:1
    $ curl dk
    curl: (6) Couldn't resolve host 'dk'
    $ ping dk
    ping: unknown host dk
    $
EDIT: Putting a . behind dk solves the problem.



working here , thanks for sharing .


looks like overstock got http://o.co for $350,000 .. is that the ballpark price for a 1 character .co?


Are they required to report how much they paid for the domain? I'm interested in seeing numbers if anyone has them.


$19,329


According to the .co registry, e.co was sold for $81,000, so I doubt a.co would be that much cheaper.


Source?


His username is "thelegit". What more do you need?!

Seriously though. I highly doubt that figure is correct.


His moniker says it all... :) hehe


text'd my pal over at amazon HQ


and he was all like "19 grand, brah!"?


lol - not gonna lie I may or may not be making fun of the question here... there's no way Amazon would ever release that kind of information.


http://z.cn goes to Amazon China.



both of this links "worked" for me, only one of them took me to the product directly, the other took me just to the homepage with Kindle suggestions instead of the suggestions served up by http://a.co i'm guessing that's what you meant.


I was not aware that 1 character domains where possible until now. I wonder why they are restricted in most main TLDs.


IANA blocked registration of most of them under gTLDs (.COM, .NET, etc.) in 1993, on the thought they may be useful to support extensibility of the DNS down the track. Country-code operators have typically followed that lead as good practice.

As they were never really used for that purpose, slowly but surely they have been released into circulation. The applications by registry operators to lift the one- and two-letter domain restrictions are at http://www.icann.org/en/resources/registries/rsep


You can have zero character domain names. Briefly trying out a few of the country TLDs, the one for Anguilla works: http://ai. the trialing dot is not needed but helps tell your browser not to try to expand it further.

Many TLDs do not resolve, though many others resolve to the TLD's registrar's site.


A TLD can have an A record, so in theory you could have a "0 character" domain. You could also use it for email, so a@co would be a valid email address if co. had an A or MX record associated with it.


I guess it's all about the money. The fact the amazon could afford it and I nor you couldn't explains it better, I guess.


I have always wondered about x.com and how they got that domain. Does anyone around here know?


They registered it just before Postrel stopped handing those out. X11 has X.org.


I'm not sure about any others except Google uses g.co for shortening their own URLs.

Edit: And Twitter uses t.co


Twitter uses t.co


z.com used to go to a site for Nissan's Z




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: