Why not discard top level domains (TLD)?(blog.pedro.si)
blog.pedro.si
Why not discard top level domains (TLD)?
http://blog.pedro.si/why-not-forget-about-top-level-domains-tld
8 comments
It isn't, he's from Slovenia. :)
And he didn't know that the TLD for his own country is .si and he never heard anything about any of his European neighboring counties TLD's? Some people simply should not blog...
I have the impression the author does not understand a lot about TLD, so his proposal isn't well thought through either.
First, TLD cannot be changed freely but are fixed, mostly assigned to countries.
Second, we now do have TLDs and I don't see a way how we could get rid of them. 'kinder.com' may be a US website promoting friendly neighborhoods, but 'kinder.de' may be a German website about children. What should 'kinder' be?
The only change I would encourage but which I don't see happening is having the TLD at the front: http://com.ycombinator.news just makes more sense (at least to me).
First, TLD cannot be changed freely but are fixed, mostly assigned to countries.
Second, we now do have TLDs and I don't see a way how we could get rid of them. 'kinder.com' may be a US website promoting friendly neighborhoods, but 'kinder.de' may be a German website about children. What should 'kinder' be?
The only change I would encourage but which I don't see happening is having the TLD at the front: http://com.ycombinator.news just makes more sense (at least to me).
The predecessor to the Internet in the UK in the 1980s - JANET - worked like this:
For example, the University of Cambridge had the NRS name UK.AC.CAM, whereas its DNS domain is cam.ac.uk. All NRS names had both a standard (long) and abbreviated (up to 18 characters) form. For example, UK.AC.CAMBRIDGE was the less widely used standard equivalent of the abbreviated name UK.AC.CAM.
http://en.wikipedia.org/wiki/JANET_NRS
For example, the University of Cambridge had the NRS name UK.AC.CAM, whereas its DNS domain is cam.ac.uk. All NRS names had both a standard (long) and abbreviated (up to 18 characters) form. For example, UK.AC.CAMBRIDGE was the less widely used standard equivalent of the abbreviated name UK.AC.CAM.
http://en.wikipedia.org/wiki/JANET_NRS
I've noticed an increase in the use of customized and
abstract top level domain (TLD) names, such as bit.ly,
bu.mp, babyli.st, etc. Even Microsoft recently
registered bi.ng
Should we get rid of namespaces and packages, while we're at it?
I'm interested to understand why you would propose removing the ".com" before first proposing the removal of "www."? From my perspective the "more superfluous of the two" is the "www." in this example.
If you haven't noticed, many website urls don't include the www.
I agree with you. My point was that in http://blog.pedro.si/why-not-forget-about-top-level-domains-... the "www" is present in the example, yet the focus is on the "com". I am curious to understand the though process that directed focus to eliminating "com" before "www"
It's simply because with most urls, when you type them without www they will work and send you to the page you are looking for, even if the url includes www. Therefore www presents no additional typing for the user. However the TLD does.
The observation you make depends on the DNS configuration for any particular domain, but ok, lets not worry further about that.
Assuming that we wish to start again with a flat name space, I'd be interested to understand:
What rules would be put in place to govern the names that would be allowed? Do you believe you should still be entitled to the domain “pedro”? Or are you willing to accept Pedro-Kostelelec-DateOfBirth ? Or some other (likely more complex combination) that is a unique distinguish-er for you globally?
Do you believe there should only be one company globally with any given name? Or is it acceptable to have a “BurgerKing” company in one country that operates restaurants, and another “BurgerKing” in the same / different country that makes BBQ's? How do you allow for those to coexist, noting the rules we already decided on in the previous question?
Personally I don't believe having a flat or hierarchical domain name space makes any difference to the underlying challenge: users need to enter the exact domain name of the business they wish to visit, if they fail to do this the result may not be what they expect. The fact that so many people choose to type some arbitrary text into a search engine and visit what ever site is returned, is fundamentally an error of understanding that should be corrected. I don't believe working around this misunderstanding by modifying the name space used serves any useful purpose in the long run.
Assuming that we wish to start again with a flat name space, I'd be interested to understand:
What rules would be put in place to govern the names that would be allowed? Do you believe you should still be entitled to the domain “pedro”? Or are you willing to accept Pedro-Kostelelec-DateOfBirth ? Or some other (likely more complex combination) that is a unique distinguish-er for you globally?
Do you believe there should only be one company globally with any given name? Or is it acceptable to have a “BurgerKing” company in one country that operates restaurants, and another “BurgerKing” in the same / different country that makes BBQ's? How do you allow for those to coexist, noting the rules we already decided on in the previous question?
Personally I don't believe having a flat or hierarchical domain name space makes any difference to the underlying challenge: users need to enter the exact domain name of the business they wish to visit, if they fail to do this the result may not be what they expect. The fact that so many people choose to type some arbitrary text into a search engine and visit what ever site is returned, is fundamentally an error of understanding that should be corrected. I don't believe working around this misunderstanding by modifying the name space used serves any useful purpose in the long run.
no.
I kind of get the feeling this proposal is as well thought out as when people wanted to get rid of hierarchical file systems in the early 00s.
The top level domains provide some organization behind domains. it also lends some authority to the system as a whole. would it even be posible to have dnssec without a tld to be signed? how would you know who to ask for information about the domain at all?
It'd be like telling all the national registrars they didn't matter anymore. especially if there's a new nation and they can't have a domain because squatters have taken over all two letter domains.
I'm sure there are a lot of other reasons why this is a bad idea, but i'm not versed enough in dns to come up with more specifics.
I feel like a jerk being this negative. but please don't start throwing out ideas for major changes in such vital infrastructure without a solid background working with it. you could get people started on some crusade that does more irreparable damage than good
The top level domains provide some organization behind domains. it also lends some authority to the system as a whole. would it even be posible to have dnssec without a tld to be signed? how would you know who to ask for information about the domain at all?
It'd be like telling all the national registrars they didn't matter anymore. especially if there's a new nation and they can't have a domain because squatters have taken over all two letter domains.
I'm sure there are a lot of other reasons why this is a bad idea, but i'm not versed enough in dns to come up with more specifics.
I feel like a jerk being this negative. but please don't start throwing out ideas for major changes in such vital infrastructure without a solid background working with it. you could get people started on some crusade that does more irreparable damage than good
Say you have a company FOO and you register foo.com. Nothing stops people in other people from registering foo.de, foo.fr, foo.tv, foo.biz, foo.info and so on, so you have to control all of these domains to protect you from domain squatters.
My proposal is to to get rid of every TLD except for .com (or something shorter) for all "normal" sites with little regulation and keep .mil, .gov, .edu and national TLDs under governmental authority.
My proposal is to to get rid of every TLD except for .com (or something shorter) for all "normal" sites with little regulation and keep .mil, .gov, .edu and national TLDs under governmental authority.
Nothing stops people in other people from registering foo.de, foo.fr ...
For example, it makes the internet less safe - people remember just the company name foo, foo.x is the real site, somebody enters foo.y and lands on a scam site
and this is a problem why? i can see taking issue with allowing unicode tlds. where some russian characters are almost identical to a latin based char-set for instance, but if you can't tell the difference between bankofamerica.com and bankofamerica.cn, you probably also fall for a lot of other scams offline too.
What about multinational companies like paypal?
Would you recommend a non-internet-savvy user from china to rather go to paypal.cn or to paypal.com/cn ?
What would a non-savvy user normally do?
In this case paypal.cn redirects to paypal.com/cn, but would you have known?
Another example: Python.com redirected to a hardcore porn site for years. Not that there is anything wrong with that, but I think both sites and their respective audiences would have been better off if there wasn't .com, .net and .org
Would you recommend a non-internet-savvy user from china to rather go to paypal.cn or to paypal.com/cn ?
What would a non-savvy user normally do?
In this case paypal.cn redirects to paypal.com/cn, but would you have known?
Another example: Python.com redirected to a hardcore porn site for years. Not that there is anything wrong with that, but I think both sites and their respective audiences would have been better off if there wasn't .com, .net and .org
Python.com redirected to a hardcore porn site for years
And which would have been more "legit" in the eyes of the public?
The homepage of an obscure tool related to software developers, or something actually related to pythons?
I think there isn't going to be a system where the domain name points to the site that is more "legit" or "relevant", that is the job of search engines. Laypersons already use google instead of dns and enter "facebook" or "facebook.com" because they 1) don't have to remember the TLD and 2) get to facebook even when they made a typo.
So a domain name should probably still belong to whomever registered it first. I'm not a DNS wizard but using whois I figure python.com was registered one year before python.org, which would have become python-lang.com or something more descriptive if there weren't these other suffixes.
So a domain name should probably still belong to whomever registered it first. I'm not a DNS wizard but using whois I figure python.com was registered one year before python.org, which would have become python-lang.com or something more descriptive if there weren't these other suffixes.
Except python.org is so much better than python-lang.com that isn't even funny.
Case in point ... I searched for "ruby language" dozens of times before learning "ruby-lang.org". And the ".org" suffix was the easy part because my brain associates those domains with non-profit stuff.
I also find TLDs to be useful for figuring out the target of a website. E.g. ".co.uk" / ".de" / ".eu" are usually companies with headquarters in the European Union.
Case in point ... I searched for "ruby language" dozens of times before learning "ruby-lang.org". And the ".org" suffix was the easy part because my brain associates those domains with non-profit stuff.
I also find TLDs to be useful for figuring out the target of a website. E.g. ".co.uk" / ".de" / ".eu" are usually companies with headquarters in the European Union.
note my point about squatters was specifically that allowing random tlds to registered, by anyone who wanted to at all, would prevent new national tlds.
I think the problem is that it would break existing network infrastructure.
You need a last component to the domain name to signify that the ip you want to resolve is outside of the local network. Imagine what would happen if somebody would register "localhost" or "main-server"
You need a last component to the domain name to signify that the ip you want to resolve is outside of the local network. Imagine what would happen if somebody would register "localhost" or "main-server"
You got a pretty much totalitarian world view. Don't you?
In most countries it's perfectly legal to have 2 (or more) businesses use the same name - provided they don't compete in the same field. In German a Microsoft that deals in software and a Microsoft that produces ice cream can co-exist.
So why shouldn't be it legal for someone in Germany to call his company FOO too? Just because you registered your name in Idaho doesn't mean that a guy in Japan is forbidden to use that name.
In most countries it's perfectly legal to have 2 (or more) businesses use the same name - provided they don't compete in the same field. In German a Microsoft that deals in software and a Microsoft that produces ice cream can co-exist.
So why shouldn't be it legal for someone in Germany to call his company FOO too? Just because you registered your name in Idaho doesn't mean that a guy in Japan is forbidden to use that name.
Sure, we can allow that, but it makes the internet more confusing and less safe. When anyone can register microsoft.x, there is the legitimate use case for the ice-cream manufacturer microsoft, but more likely it will be a scammer or a phisher or something like that.
If we give up these nationalized namespaces (which aren't really enforced anyway, see .tv and and .ly etc) then we have one canoncial name for normal web sites, and microsoft ice cream could be found at microsofticecream.com
By the way, I don't mean "allow" in a legal sense. Everybody can run their own DNS system and be happy with it, I use the word in a technical sense. Everybody can call his company how he prefers, this is about DNS.
If we give up these nationalized namespaces (which aren't really enforced anyway, see .tv and and .ly etc) then we have one canoncial name for normal web sites, and microsoft ice cream could be found at microsofticecream.com
By the way, I don't mean "allow" in a legal sense. Everybody can run their own DNS system and be happy with it, I use the word in a technical sense. Everybody can call his company how he prefers, this is about DNS.
> Sure, we can allow that, but it makes the internet more confusing and less safe. When anyone can register microsoft.x, there is the legitimate use case for the ice-cream manufacturer microsoft, but more likely it will be a scammer or a phisher or something like that.
So we should give up our freedom because there are bad guys abusing that freedom?
So we should give up our freedom because there are bad guys abusing that freedom?
Erm, is this a troll, a linkbait or simply ignorance? Those "abstract" TLDs belong to real countries. (Well except your .42 - which is a custom TLD that's not accessible through the "official" DNS system.)
First, there's no such thing as an "abstract TLD". The closest analog would be generic TLDs. The author actually manages to get this completely backwards since the gTLDs are .com, .net, .org and their brethren. The TLDs the author calls "abstract" are all actually country-code TLDs (ccTLDs), and they all have a proper meaning. .ly is Libya, .st is São Tomé and Príncipe, and .ng is Nigeria.
Just because these governments don't traditionally care too much about who uses them doesn't mean they don't have a meaning. They aren't abstract.
The address bar in Firefox will actually redirect you to google.com if you type just google, however what I am proposing is to actually get rid of the TLD for some websites.
This is pure fiction. What actually happens is if Firefox can't resolve the address, it assumes it's a search term and passes the term into a preset URL (defined by the keyword.URL setting).
Back before version 3.0 (I think; don't quote me on that), this preset passed the "I'm Feeling Lucky" flag to the Google search URL. As a result, if you typed "Google" into the address bar, with no TLD, Firefox would effectively search Google for the term "Google" and the "I'm feeling lucky" feature would redirect you to the first result, which is obviously Google.com.
Since 3.0, the preset doesn't have the "I'm feeling lucky" parameter, so if you type "google" into the address bar, you'll actually get a page of google results for the term "google". I, and I'm sure many others, have since tweaked the keyword.URL setting to again use the lucky flag, but on a vanilla install, the author is just plain wrong in his assumption.
One final nitpick:
* After this recent surge of custom TLDs, I am beginning to think, that they are not a limitation, but that we can change them, almost freely, presumed we have the knowledge to do it (Please, correct me if I am wrong, and forgive my ignorance).*
Perhaps English isn't the authors first language, and if that's true then I apologize, but I simply was unable to parse that paragraph. Commas just don't work that way.