Internet Metadata - Open Collaborative Rating 71
XenonOfArcticus writes "INMD is a recently-submitted proposal for an open collaborative profile-based ratings system for Internet Web/Usenet content. Based on a system partially inspired by Slashdot itself, it seeks to facilitate content rating that does not force censorship. The purpose behind INMD is to allow the user to choose what criteria to use when filtering/sorting content, and whose opinions should impact their browsing.Don't want the RIAA or MPAA to influence your Internet? Exclude them. Want your opinions controlled by the NRA? Submit your browsing to their discretion. You remain in control. " Looks still quite beta, but it's a cool idea - something like everything, in terms of rating knowledge/content.
on dear.. (Score:1)
asinus sum et eo superbio
Great idea, but hard to do on a large scale. (Score:2)
Slashdot works becouse there is a relatively small amount of content, and a large user base. Reverse that, and you get someones great idea that never comes to fruition.
On a similar tack.. (Score:2)
Re:on dear.. (Score:1)
-awc
Re:on dear.. (Score:2)
To make things more manageable. Look at usenet, it is currently an extreme case of "information overload". There's quite a lot of good/interesting stuff in it, but sometimes it's like sifting cubic meters of sea water to filter out gold atoms: not worth it.
Information overload is IMO becoming a real problem. Collaborative rating might provide a way to deal with it.
Re:on dear.. (Score:1)
I, however, don't know how well it will work for entire websites... will it rate pages or machines, or entire domains?
ap
This could actually be good... (Score:1)
Anyways, think of it this way, parents scream for security from the evils of the 'net for their children, well, this way you can say, look, this page was judged appropriate by [insert company/organization name/random person here].
While this still goes by others opinions...you can choose who's opinion you'll go by...some may be good, some may be bad (i.e. Disney deeming Britney Speares half-nekkidness appropriate to hit kids with...but that's another story for another time).
Good help deal with spam, and while the article seemed to focus mainly on NNTP, it could work...maybe.
I wouldn't use it, but hell, it'll shut up the parents...and after all, isn't that what we really want?
Who do you trust? (Score:2)
In the more "traditional" media outlets, we've had little to no choice as to what that filtering mechanism has been. Going to alternate sources (such as the internet!) has been an iffy exercise -- much more information available but without the same credibilty as the "safer" outlets. (Whether the credibility or lack thereof is actually deserved is a whole other topic...)
A system such as this allows people to choose who they trust to act as filters for them. The question comes down to who it is that you trust to make these descisions for you. There's a loss of personal freedom in ceding your though process to others, but a large gain in convenience.
Of course, a good filtering system should let you turn it off whenever you want...
--
A few musings (Score:2)
Also, it might be a good idea to have some kind of "matching" system whereby your taste (in raters or items) is matched to other's so that their opinions can be taken into account. This is somewhat similar a recommendation service I've seen on amazon [amazon.com]. I haven't studied GroupLens [umn.edu] in detail, but perhaps it has these aspects too.
Re:Morons ranking morons (Score:1)
Try looking at someone's User page who you think is making a grain of truth, a good statement, or otherwise just well spoken, you may be suprised when you see someone with +100 karma (their statements tend to be consistently good), you'll see that the cream truly does rise to the top.
-[ World domination - rains.net ]-
Remeber Firefly? (Score:1)
I always thought the old "Firefly" model of collaborative selection/filtering could be useful for "Net Nanny" types of applications.
Here's where we find out whether the goal of the CDA types is really protecting children (in which case this solution is acceptable) or censorship of society at large (in which case, the freedom allowed by this scheme would be unacceptable).
Bets, anyone?
Also, apropos nothing, did you notice the file sizes for Draft 01?
Text: 11K
HTML: 16K
Word: 177K
Yes, Word requires a tad over 16 bytes per character of text. (Yeah, I know, it would be a lower ratio in a larger file.)
Categorization (Score:2)
Another issue is the large data / small readership problem brought up in a previous post. Unless organizations are willing to go and rate site upon site, how will there ever be enough rating information to be useful? Rating books and albums works on Amazon.com partially because there's a simple 5-star customer review system (even simpler than categorization) and partially because it's a restricted data domain. Also, if someone writes a review for one book they've read, they're likely to write reviews for others.
Maybe web sites should offer a "Rate Me" button linked to some third-party rating company. Inspired surfers can follow the link and rate the site on a few static categories or offer their own comments. This could also give us cross-information of the "surfers who liked this site also liked
The main problem with this models is most sites won't offer a "Rate Me!" button. Alternatively, a rating company could set up a "meta-site" in which you can surf (using frames, for example) and have a "Rate This Site" button in the margins. Unfortunately, rating by URL is risky since the often change, but it's a first step.
Anyways, I guess my point is rating systems go beyond how they're designed; you also must consider how they're deployed (as other posters have mentioned). Solve that problem, stick an ad banner on it, and you've got the next great web fortune which will be bought out by Microsoft and eventually hacked by someone reading this.
cotholo.org, or, THE RETURN OF JONZILLA (Score:1)
This is really just kind of a silly idea, you have to pick a group to moderate your content? I'd rather moderate my own, and I can do this with my brain, and alot faster too, no group or person is going to feel EXACTLY as I do about a wide range of things, (this is the point on reading content, your view point and knowledge can change.) and personally I wouldn't even trust or have the audicity to ask Linus Torvalds to sift thru information on a subject like linux.
OTOH, continuing with the Linus example, it's not censorship if it's only blocking lies, for instance if Linus Torvalds censored out things like 'LINUX USERS KILL BABIES AND EAT PEOPLE'
But as per any ratings system, laws can be hung on it. This has been proven with the australians who hung their laws on filtering software, and barring that they just rip a list off the filtering software and put it on the ISP.
Summary: Someones view point and knowledge base can change, which is why it is a shame to have one group of people censor or otherwise promote things IF it is used as the only resource for learning about such things. Furthermore Any ratings system can have laws hung on it, including this one.
-[ World domination - rains.net ]-
Re:Morons ranking morons (Score:1)
I'm at -7
But this post will drop that to -8.
Oh well
nearly impossible to censor (Score:2)
RFC2616?
Make a little daemon that sits on port 80 and browsers work with, but doesn't fully comply with the RFC.
Anything on port 80?
Change the port.
Anything that uses HTML?
What about Flash, java and other things of the sort? Or use a cgi script to insert the contents of a normal text file in a page... the illegal content would not in HTML by nature, just inserted in it because of the viewer.
The most feasable solution is client side filtering and this is also riddled with holes. What defines a client? If you telnet to a site and read it would it be illegal? Would there have to a commitee to decide exactly what you are allowed to use as a web browser? It would either fail pitifully or turn big brother.
If all else fails, make a new protocal that is cosmetically the same but functions differently.
In short: it's near impossible to do and legal battle after legal battle would have to be fought over it wasting more money than it would ever be worth.
Re:Morons ranking morons (Score:1)
I'm obviously a troll
-[ World domination - rains.net ]-
Re:cotholo.org, or, THE RETURN OF JONZILLA (Score:1)
Personally, even checking what each thread started off being about would take me the best part of a day.
Re:Who do you trust? (Score:1)
One Weakness . . . (Score:1)
Think that this can't happen? Look at
http://technology.news.com.au/frameset/frameset
And similar attacks have been launched against groups in the news.admin.net-abuse.* hierarchy. Some of the net.kooks & spammers don't like people reporting their attacks for some reason.
Unless any rating system can distinguish between real posts & forgeries like these -- as well as cope with the distributed nature of Usenet -- this will be just another proposal that everyone considers is ``a nice idea if we could only get it to work".
Geoff
Re:Morons ranking morons (Score:2)
Slashdot is not an example of this at all. When reading Slashdot, you do not choose which moderators' opinions are used to score the articles presented to you. This collaborative stuff if far superior to Slashdot.
Ideally, if you wanted pro-Microsoft posts at the top so that you read 'em first, then you would just select a pro-MS moderator (or an averaged board of moderators, some of which might be computers that automatically mark down documents that contain "LINUX RULEZ D00D", for example.)
Er, anyway, please don't confuse this with Slashdot-style moderation. What these guys are discussing is much, much cooler.
---
Have a Sloppy day!
This would never work (Score:3)
Firstly, consider just how massive and transient data is on the Internet. Now imagine for yourself a database capable of indexing and storing *ratings* for this massive volume of data, including necessary security precautions such as "this-person-has-already-rated-this-item", etc. The amount of data required for such a database is truly staggaring, and, depending on the popularity of an item being rated, could likely exceed the storage requirements of the item itself. This isn't even considering the database of users that would be required in an authenticated setting as he's suggesting.
Assuming one could construct such a titanic database, a rating system like this would only really be effective for static content (like non-dynamic web sites). Things like newsgroup articles stick around for a few days -- weeks at most. This is hardly enough time to get "moderators" to accurately rate articles in all but the most popular newsgroups. Face it: Slashdot has several orders of magnitude more readers and posters than any one newsgroup heirarchy in existence. It would be a rare thing to have more than one or two posts in a given newsgroup rated, and that rating would be virtually meaningless because of this.
The bottom line is that this is unfeasible and would not be adopted widely, which would be essential for its growth and continued use.
Just my $0.02 ($0.02 Canadian).
Worth looking into... but (Score:2)
The most serious concern is that a government could somehow piggyback on the system to implement effective censorship.
The issue of upstream filtering does not go away.
As described, this may just shift the battle from which pages should be banned, to which ratings groups should be banned.
As pointed out, many technical issues like bandwidth are involved.
People being what they are, it might be far more common for people to moderate against pages that offend them, than for good pages to be promoted. Remember, people don't vote for important things, like national elections, and this is webpages.
Should this work like Slashdot, with only one group of raters chosen randomly, or by various groups with an agenda you can choose between (or both)? In the former case, you need attributes like Informative, Insightful, Troll, except they probably read more like Sexually Explicit, Rude Language, Violence, and Politically Subversive. Note those are all "negative". Finding a positive attribute set that would promote the right pages to the right people seems difficult. Also, summing attributes simply won't work, you need a complex filter that weighs the entire set thru a scheme controlled by the user. (Hey, this would be nice for slashdot too!)
It also leads to something of a privacy question in that it becomes trivial to determine a web publishers political views, by looking at their rating spectrum.
I think the authors should be encouraged to complete their proposal, and given feedback by all those concerned by the drawbacks. *If* a grass roots rating system that was impossible to abuse could be designed, it would be solution to a very important problem.
I have to admit you may have a point here. (Score:1)
For determining if something is worth my time to read it usually takes me about five seconds to infer whether the post is a flame or a friend.
-[ World domination - rains.net ]-
Great feedback so far (Score:1)
Don't view INMD as censorship. It's not. Or, doesn't have to be.
It can be used to censor things by rating them negatively, and if you choose to allow yourself to be directed in that way, by "authorities" who will shield you from things they think you don't want to see, then that is your choice.
I would prefer to use it in a sorting manner, and would choose rating-groups that I believed would mark good stuff as good, and ignore the crap. In this way, I can choose to see _everything_, but look through the good stuff first.
Heck, you could use it as anti-censorship. Form your own smutpr0n ratings cabal, and go seek out the best-looking sheep on the Internet, and let the world know where to find 'em.
It's just a tool. It's a more flexible tool than some of the systems out there, because it can be used for good, not just for censorship.
And yes, the Word document is a pig. My apologies, but my TeX is very rusty.
Re:This would never work (Score:2)
Take some technologies like WebDAV for dynamic publication of web content, defini a DTD for a "Web Rating" for use with XML content, and then figure out some method of scripting that would allow a person browsing a web page to simply click the "Rate this page" button which would send a new "Rating" to the WebDAV engine, which could then parse it and update whatever the particular XML content of "Current Rating" to average the new rating in.
Of course, this is really really not thought through at all, but I think it's at least possible, if not exactly feasable. It would require that all websites supported something like WebDAV and scripting and the proper XML stuff online for one thing. But you get the idea.
-=-=-=-=-
Re:Who do you trust? (Score:1)
Uhm, NO. (Was Re:Morons ranking morons) (Score:2)
That's completely false.
Every day I expicitly verify that some 10 randomly selected moderations were fair. Many days I find 10 fair moderations. Only occasionally do I find one I thing was unfair.
Anyway, the comments that are moderated up include many viewpoints, many of which don't mention Linux at all. There are even some comments that get moderated up which point out faults with Linux.
But then, if you were involved with /., you'd know that. In case it wasn't already clear, you are welcome to get an account here and participate in the meta-moderation. You could even attach a username to the things you say.
Re:This would never work (Score:1)
Re:cotholo.org, or, THE RETURN OF JONZILLA (Score:1)
My impression of this system would be like you and a group of your friends splitting up the 'net and everyone searching for things the group was interested in and sending them to each other. Everyone would search a different 'part' of the net or for different things. Not censorship per se, though you would lose out on some content in some places. Heck, come to think of it that's the best idea I've had in a few days, I think I'll go tell all of my friends to start e-mailing me things that are interesting.>:)
Kintanon
What happens if ... (Score:1)
If I submit my browsing to the RIAA, I guess I won't be viewing many mp3 sites. Similarly, if I submit my browsing to Linus Torvalds, I guess I'll be viewing a lot of Linux sites, but not many Microsoft ones.
If I submit my browsing to Anonymous Coward ...
Just a quick guess at storage requirements.. (Score:2)
The "draft" being discussed calls more for an independent collection of servers holding and serving ratings information. It seems the storage requirements for such a thing would be enormous enough, but I just now started thinking about the *bandwidth* requirements. For every web page you request, for every newsgroup you download, a connection must be made and ratings retrieved for everything. Yeoww..
Re:cotholo.org, or, THE RETURN OF JONZILLA (Score:1)
I hate to burst your bubble but you've basically just described what
-[ World domination - rains.net ]-
Re:I have to admit you may have a point here. (Score:1)
Ok then, let's start with a nice low estimate and say that there are 1 billion distinct individual pages making up the entire internet at this moment with none being added.
That's 5 billion seconds... Lemme do some quick math here, 5 billion/60 = 83333333.3333 minutes = 1388888.8888889 hours = 57870 days = 158 years...
Looks like you'll be dead before you get half way there most likely. Especially since that's 158 years without doing anything else, if you only get say... 8 hours per day then it's 3 times as long... Of course, if you throw 10000 people at it you can cut that down to just a few years, if they do it full time, and no knew content is added.
But in the meantime, I hope you don't have anything better to do.
Kintanon
PICS Contrasted (Score:1)
I prefer to let content authors choose their tags. If you foster the emergence of trusted authorities of any description to judge site content, you open the door to rating based self-censorship far more than the more modest PICS based approach. If the site publisher chooses his or her own tags, they have a choice about which classification schema and which tags they deem are appropriate. The feedback mechanism is controlled by the publisher and can only be abused to the extent that the publisher can be coerced.
In this system ratings are polls. They carry far more weight and are far more useful for censorship purposes. The damning judgment of grandstanding politicians will hold sway over public internet outlets like schools and libraries. The same bombastic crusaders will be given a potent tool to pass judgment on publishers. These judgment will inevitably take on a life of their own and will be used as a weapon against 'offensive' sites. If Gary Bauer's ratings gain a following, that following will be used not just to influence his flock, but to coerce a site's advertisers and associates. They could extend beyond the web and affect the publisher's non-virtual life too.
USENET has died in a cacophony of idiots, may it rest in noisy, unrated peace. At least I can still pull some good MP3s out of the dreck. Moderated newsgroups and interactive, moderated or edited sites like /. are filling the void nicely. I don't really see the value in improving USENET's S/N. If you succeed, do we really gain anything?
CritLink (Score:1)
CritLink [crit.org] provides free (free source too!), universal web page annotation without client-side software. The annotations appear inline with the webpage when viewed through CritLink, and can be stored on your own webpage rather than on Crit's servers.
Do we have this already? (Score:1)
I have my content filtered through Slashdot. Yes, that's right. I don't have time to search the web for interesting news, so I let the like minded people here help me.
The NRA or RIAA also have websites where I can find information that each think is important. Yahoo also has something similar, but they call theirs a 'portal'.
This proposal seems to be an answer to a problem that is already solved. It boils down to a proposal to wrap-up all of the existing portals into one big portal.
Re:I have to admit you may have a point here. (Score:1)
As for spending 158 years looking at web pages this is just a huge 'what-if' which, in another form is a useless factoid. I'd only look at pages that I was intrested in the first place via search engine, (which I shall remind you this is NOT, this is, essentially a site where they review pages.)
Do you look at every book in the libary when you go in? Of course not, you use the card catalog. When it comes down to it, I can decide what i'm intrested easily, and we all know this. I'm not going to look at web pages on say David Bowie if I'm not intrested. I'm not even going to venture close to them. Ultimately although this site may have some uses it will probaly end up having more cons than pros.
-[ World domination - rains.net ]-
Re:On a similar tack.. (Score:2)
However, both subject classification and call numbering (and the assignment of keywords and descriptors from controlled vocabularies) are not easy things to do. I've taken three master's level courses on cataloguing, classification and indexing, and I won't call myself an expert; I don't expect the average Web or Usenet author to put even a fraction of that amount of time and effor into it.
Cataloguers often spend quite a bit of time learning a classification system and how to apply it correctly (meaning consistently with the way everyone else does it, among other things), and for the most part, cataloguers specialize in a particular subject area. Even then, it's rarely a five minute job to properly classify something.
The inherent problem is that we have the single largest repository of information (using the term loosely) and we're essentially leaving it up to people who by and large have no cataloguing, indexing or abstracting skills to independently look after their own small corner of the mess and hope that somehow a universally workable system emerges. Ultimately, both Usenet and the Web will probably prove to be unorganizable in any universal way, and the best way to find material that you find useful is to determine what other people like you found useful.
Anti-ratings (Score:1)
Just a passing thought... You say:
Also, in your proposal, you mention using other people's filters so you would see what the NRA deems good (or whatever). Now, put these two ideas together, along with concerned parents... How about anti-filters, where you decide to block anything which the porn cabal marks as "good"? Seems like a natural step in which to take things, but I do not see it explicitly mentioned in the proposal.
Mix In, Mix Out (Score:1)
Never work?! HA! (Score:2)
The key is, of course, allowing the user to filter ratings from any person/group they disapprove of.. or also allow composite scores from several groups (say, for example, you want Rush Limbaugh and Newt Gingrich to edit content for you... Rush rates 'Hot Sexy Babe Website' absolute filth, while Newt maybe only rates it 'indecent'. Composite score: 'Worth a Visit').
The key here is to allow the user maximum flexibility with the system while ensuring that only people who are going to make an honest effort in the system gain access. Nominations are a time-tested method because it simply relies on peer pressure ("I nominated you - don't embarrass me now by acting like an ass!").
So there. Oh, and Rob.. if you're listening... how about we make that long-awaited revision to the current moderation system and open it up alittle (like the suggestion above?)... it's time to try something new...
--
I've got an idea (Score:1)
Beer recipe: free! #Source
Cold pints: $2 #Product
Somebody is already doing this (Score:1)
Re:Morons ranking morons (Score:1)
Some thoughts. (Score:1)
"...but on a larger scale, it only draws attention to the posts that happen to get rated high..."
The answer IMO, is to default everything to a 10
out of 10, and then filter things down.Don't
consider everything junk unless rated high. Instead, consider everything great, unless rated
down.And this should work, because:
a)There would be a lot of people rating.
b)If you rate down a user's selection, you rate
down ALL his/her selections on that category.
BTW, IMO, there shouldn't be URLS/article ids
rated, but rather persons.(Authenticated with
unique encrypted ids.)
So you wouldn't have to say:
"I like/dislike www.somewhere.there",
but rather:
"I like/dislike user X's selections about
the category, ie., politics/economics".
That way there can also be formed groups of people
with same interests.Take notice, that I say
"interests" and not "views", because mayby I
don't agree with user Y about politics/economics,
but I find his ideas interesting, and so I want to
read what s/he has to say.
Finally, I think that the most difficult part
will be to come up with these categories.
I mean, is it politics/economics, or
economics/politics, or ever economics/applied?
(For a practical experiment, have you ever tried
to locate a not-so-obvious setting in a
submenu of a submenu of a submenu, ad infinitum,
of emacs?(Thank God for apropos))
Filakia
Re:This would never work (Score:1)
You would not need a central store of ratings, either. Major organizations would publish their ratings via their own servers. Individuals could publish via Yahoo!, their personal servers, IM client, etc.
Re:Anti-ratings (Score:1)
Want to reclaim the glory days of Usenet? (Score:1)
Well, it's worth a shot.
--- Dirtside
Possible solutions to concerns? (Score:1)
1) The internet is too large
2) The system is a form of self censorship in that you moderate out opinions you do not agree with.
3) Who do you trust?
4) Moderation anomolies.
5) Normal human differences (i.e. no two people have exactly the same opinions.)
6) Groups modifying the code to favour one view over another.
7) Hanging censorship laws on it
Relying on just profiles moderation leaves the system open for abuse and censorship. I think the addition of a profiless rankings system will be needed as a balance to the profile rating. I think that using a statistical-neural approach will probably be the most successful.
As far I have thought it out at the moment, it works like this. The ratings would need to be split into five different categories general, news, entertainment, science and technology and arts. All information units would have a general rating. Each site then falls into one of the other three categories.
Statistical calculations are carried out on the ratings for each post. Most posts should show a guassian distribution of ratings around a point. I think that most anomolies will show up as spikes at either end of the distribution. To reduce the effect of these on the post ranking, ratings outside say one standard deviation on either side are weighted downwards.
The neural network comes into play by looking for patterns in posts. It examines posts with ratings at several key points suchs as 1/4, 1/3, 1/2, 2/3 and 3/4 (Note: I think 0-10 or -10 to 10 will not be large enough. I think -100 to 100 or 0-100 will be needed) of the range. With in these posts the neural code is looking for information patterns that characterise posts at those particular points. It then uses this information to rate new posts. This rating can then be used as the starting point. It also provides another check for anomolous ratings. If something is really way off then the neural network can flag it and/or after a comparison with past posts add in weighting factors. To continuely improve the neural net, its' ratings and the public ratings are compared. The neural net would also continuely analyze new posts at the points to better characterise them.
The neural net is not looking at the information per se but the characteristics of the post. I'm not sure are this stage what the characteristics would be. There was an article in the New Scientist from this year that looked at the research by several people into detecting when a report is missing pertinent information. I was thinking that a similar technique as their's could be used to characterise the posts.
In effect you have three ranking systems the public, the neural and the profile ranking. Once the neural has had time to operate it will help to classify all those pesky dynamics of the internet. Note that the profile ratings can also be statistically examined and inputed into the neural net. A generalise profile rating would supply a fourth ranking.
One major concern is somebody(people) changing the code to weigh particular information up or down. The best solution to this problem is open source. It would be very hard, if not near impossible, to slip code favoring information in without finding out. If they did it would almost certainly be found out and the reprecussions for those people would be devastating.
The internet is large and this will definetly be a problem. One solution is two use many distributed ranking servers. This would also provide backup. Based on the draft, I foresee that you would have one server as a database recording the rankings (public, profile, neural) and message id/URL. Another server does the statistical calculations and runs the neural net. Distributed servers would provide an elegant way of chewing through the large amount of information and in the case of the neural net provide a rapid development path.
As to the last problem, well any system either governmental or public has the possibility of laws being hung off it. The realty is if we don't come with a reasonable rating system, vested interests will create their own. Now which is worse?
Simon.
Link w/o javascript (Score:1)
The link posted above requires javascript to be enabled. This is unstable in many Linux versions of Netscape.
This [news.com.au] is a direct URL
Re:Link w/o javascript (Score:1)
Mea culpa. This poster will be more careful in the future.
Geoff