Well I don’t have anything more to offer as a counter to the ideas presented here except this:
Instead of a defacto NRS, my preference would be for a defacto SEARCH - in-built Maidsafe search. Something ajaxy that pulls site titles (not domain names) as you type according to ranks given by whichever algo you’ve picked - maybe maidsafe would set a default one.
So if I went in and typed google.com … then I’d get a list with presumably the most accessed site at the top (most accessed can possibly be determined via how many copies SAFE is storing or some such info scraping of the network) and/or use google’s old now out-of-patent pagerank algo.
In this manner of Search, we don’t need an NRS, you just need a name and the search does the rest. It’s easy, existing companies can utilize it with no problems alongside QR-codes if they are small startups with a common name.
This would be a revolutionary approach - just as SAFE is a revolutionary network and would fix many many problems that we encounter with existing DNS and existing search on the clear-net.
And instead of ads in the search, a simple donate to the maidsafe foundation link
Oh, another thought - when we get “compute” in SAFE, then the search DB could possibly be decentralized as well. … or maybe that’s not even needed if built into core code?
I’ve been describing a proposal along these lines, but it seems to have been lost in the mire of the thread. It’ll be given even more power via linked data BTW.
We’ll still need NRS though. As I said before, it’s for communicability, not findability, and the two systems are not mutually exclusive, but work together.
QR codes are useful in some limited circumstances, but are not a solution to making XOR urls usable.
You still need something readable, constructible, understandable, memorable, pronounceable, and that gives you some idea of where it might lead.
And let’s not forget about accessibility either. XOR urls (and of course QR codes) are a horror show for people with visual impairments.
I can see some unintended consequences with using a search feature such as scam sites deliberately having their site feature highly in searches and appear to be the real deal.
A page that has links in it should be able to use the site “name” and the user doesn’t have to choose from a search list of which actual site the link refers to.
Also I should be able to put in safe://xyz and just be taken to the site with the “name” xyz
OR
Are you just meaning a way to search the “names” as a specific action and not a way for links to work?
I’m not talking about search in the Google sense of search, but a way of using say a title (aka non-unique alternative name, chose by the owner of the site or identity, not simply the HTML page title) that would be used in combination with the unique NRS Name to quickly narrow results in the address bar, in a type ahead fashion. This, in combination with some socially derived weighting (perhaps similar to pet naming), could provide a solution to get the most relevant, and trusted site to the user quickly.
An example:
I type in “microsoft” into the address bar, and this is what gets typed ahead as first result:
Microsoft | safe://microsoftofficial
Because most of my friends/contacts have it in their bookmarks, and/or it’s the top recommended site in the site community derived site verification list I subscribe too, and/or most people tag it Microsoft.
Whereas
Microsoft | safe://microsoft Microsoft | safe://micro5oft
Appear only after I tab down the list as no-one I know has booked marked these sites, and they are often tagged as fake-microsoft etc.
But, links would still work with the NRS when i’m constructing the site, or when I paste from the address bar. Or perhaps when the a pet naming / recommendation becomes strong enough I could be so bold as to just paste a “tag” or two non-unique elements of a linked data triple, and that would reliably get others to the same place, without the need for the NRS name.
The Do Nothing Approach. Let NRS names be claimed with abandon just like how alpha 2 and current test vaults operate.
Pet Names. Require plugins for the SAFE browser that enables the user to manage a local NRS of their choosing.
Safe Search. The NRS is combined with a search-like mechanism built into the browser. When someone enters “safe://cats” or “safe://happy.being” multple options are returned in a sorted order when there are more than one claims on that name/id. The user is then required to pick the real “safe://happy.being” they are interested in. Could be combined with option 2 to cache previous selections and make repeated visits less cumbersome.
The Perpetual Public Reserve and Private Squat. A public reserve of all current cleanet tlds, all single words in all online language dictionaries, and all proper names like “John” or “Scotland”, that cannot be claimed by any one entity. The keys for these public reserve names are published as version 0 of the site. Any other name not on this list (such as compound words or l33t spelling ) is available for private ownership, profiteering, or the perpetual squat.
The Great MaidSafe Auction. The same as option 4 but instead of a public burning via version 0 publishing of private keys, the tlds and common/popular public names are auctioned off to the highest bidder to generate funds or offered for sale to “valid” clear net recipients. A variety of other sale terms possible as well.
NRS Composting. Any one of the above options, but instead of having perpetual NRS, there is a time to live. If the current owner does not pay a maintenance fee to the network, then the NRS entry is reset and available for someone else to grab.
Just thinking out loud, without any technical expertise. Would it be possible to have something like this:
XOR: safe://c4cc596d7321a3054d39724ff82fe64f49c3896a07a349d31f29574ac9f56965
Default NRS resolve: safe://sascha
Name resolved by browser plugin foo: safe://sascha.foo
Name resolved by browser plugin bar: safe://sascha.bar
Maybe the “.” character could be disallowed inside the URL and “services” could be denoted by simple “/”. E.g., where (.foo) is optional:
safe://sascha/fishing(.foo)
safe://sascha/linguistics(.foo)
Developers would probably want to use XOR in the code when referencing web pages.
EDIT: Strike the above as irrelevant to the suggestion.
My suggestion is to have the option of referencing the system used for resolving XOR to human readable format in the address name. Just the default could be NRS.
Thanks @jlpell, excellent list. I would add @nevel’s::
NRS names only in address bar: NRS is implemented as in alpha2 (option 1) but only function in the browser address bar. To link inside HTML etc, you must use XOR addresses (described here).
Then you have no way to verify the link is sending you to where you think it should be. Mouse over of the link in the html page would only show a meaningless XOR address. Just like scammer do by using an IP address but show you a legit looking link text
Use something external to SAFE
Another plugin solution. Since SAFE will not have smart contact capability at launch, leverage a blockchain that does until such time that SAFE has the capabilities to implement NRS in way that provides a suitable trustless means of distributing the names. ENS already has the capabilities required/desired. No need to burden Maidsafe with any of this.
I am thinking of the average user who also checks that links on a page are what they should be. Even sites are advising visitors to do that as good security. (banking, government, antivirus, ?) sites have this advice in various places. Maybe its more an Australia thing where sites from here are giving that advice, but it is essential advice.
Yes I will repeat my opinion since its for the public we wish to attract and to have scammers able to trick people because of “normal practice” is bad practice.
Yes we need a way to be able to use the actual XOR address but it will not be good at this time to do that.
Maybe we can make the browser do a reverse lookup in the NRS to show the XOR and site name when mousing over. But at this time that is not there so XOR as the underlying link is bad practice. And yes I know we need to do that but not until we have something like the reverse lookup on mouse over
I agree. Using XOR in the code was not my main point, though. I mainly thought it might make sense to state the resolver, NRS or otherwise, as part of the address.
Yes, but once scammers know of that flaw then its worse than security by obscurity, useless
I think that the browser doing a reverse name lookup would solve the problems with having XOR addresses as the links (be textual or underlying). @JimCollinson do you know if a reverse lookup is possible to implement for mouse over of a link on a page?
