The Open GRiD Project | ||
|
| is this? | |
|
| ||
| need this new thing? | |
Potential global implications of implementing the architecture seem to be surprisingly substantial for many areas other than just Internet searching. | What are those implications? | |
The basic idea is to capture opinions of very many people about each Web page in order to find out their weighted collective opinion about the page; and then use that opinion as the main part of page description (together with its title, keywords, text, etc.) when making a Web search. That is, we will build a collaborative information filtering, classification, and ranking system. | What exactly is proposed? | |
Intuitively, the proposed method gives everyone the ability to easily express his/her opinion about anything; and to easily find out the collective expressed opinions of all the other people on the Web about anything. | What are the main benefits? | |
Also the proposed idea can be viewed as a neat way to easily use everywhere on the Internet the benefits of peer review and peer recognition (which are widely adopted e.g. in science and in Linux/free (open source) software developers community). | Tell me more about this! | |
Technically the idea develops on the methods used by the Google search engine (it can also be seen as the next step after the Open Directory). But the proposed extensions should allow for making much more flexible searches and for getting much more accurate search results and categorizations comparing to the current search engines and directories (including Google and the Open Directory). | Tell me more about this! | |
The implementation of the Open GRiD project has been started, but some collective efforts of many different people (both developers and opinion content creators), that is, your efforts, Dear Reader, are required. | What can I do? | |
|
To the Open GRiD Project Research Papers |
On the other hand, you might wish to look at the Frequently Asked Questions first: there you can quickly find answers to some of your questions, like the ones on the margin above or the following ones, without scanning through the whole document:
|
To the Frequently Asked Questions |
|
To The Open GRiD Project Architecture Proposal |
|
To the Software Download Page |
 |
Back to the main page of the Open GRiD project. |
This project is constantly evolving.
What is one of the main problems of the Internet today?
The problem with most current Internet search engines
is that the search results
are based *only* (mostly) on the contents of the page search engine
gives you as the result of your query.
(Categorizing directories like the Open Directory
-former NewHoo-
(http://dmoz.org/)
and Yahoo!
(http://www.yahoo.com/)
are discussed in the next section.)
The above searching principle works exactly as good as the honesty
and more importantly
just the knowledge of each person that set up a particular Web page goes.
If the author describes his/her Web page (by its title, keywords, text, etc.)
correctly
and accurately (with respect to the methods used by the search engines)
you can find his/her Web page using current search engines.
But you will not lose his/her page among hundreds of other search results
a search engine gives you only if *all* (a sufficient majority of) Web authors
describe their pages accurately.
What makes things even worse is that HTML does not have almost any
facilities designed to help search information on the Web
(it was designed simply as a language instructing a browser
how to view the information).
Obviously, no search engine (or a single firm/organization/country) can
enforce that all Web authors describe their pages accurately.
This is just
impossible because of the limited knowledge of each Web author even if
each of them is absolutely honest and does his/her best.
For more information on how search engines work and select the pages
to display first
see http://searchenginewatch.com/webmasters/work.html
and http://searchenginewatch.com/webmasters/rank.html.
Also, it is impossible in principle for any search engine
to extract accurate
information about a Web page looking solely at what is written
on the page itself (which is what most of search engines do):
For that the search engine must know *everything* and be able to apply
this knowledge to judge relevance/quality/etc. of the Web page.
Just recall the frustration you frequently get when trying to find
something non trivial using current search engines:
You get hundreds or thousands search results most of which are
completely unrelated to what you were looking for.
Now imagine a much better situation: the one when you can utilize
all the brain power and time
of thousands of previous surfers to advise/recommend you
which pages are (more) relevant
and provide (better) information/services/etc. you are looking for.
Read on to find out how we can make it become a reality.
Another point is that current search engines can be pretty easily fooled:
Remember those XXX porn sites popping up in first 50-100 search
results when you
search for something vague and popular? Such sites just put a bunch of popular
search words and phrases into the fields describing the page to a search
engine to fool the search engine.
Part of subsection
"How the Proposed Solution is Related to What We Know"
later in this document describes the relations of the proposed solution
to the approaches described next.
The best practical attempt to solve the Web searching
problem that I am aware of has been so far
the Google search engine
(http://www.google.com/).
Their approach
(http://www.google.com/more.html
and http://google.stanford.edu/google_papers.html)
is shortly as follows:
Each Web page has a rank.
The rank is high if many pages with high ranks contain
links to that page.
This is a recursive (cyclic) definition, so they just roughly
speaking download all this information about very many pages and
links on the Internet
and then process it as a whole to compute those ranks.
You can check out this search engine
(http://www.google.com/)
yourself and see that it really works, what
kind of results it gives, and what ranks different pages have.
Supposedly the higher the Google rank of a page the more valuable,
high-quality information
that page should contain.
The intuition is that pages referred from "good"
pages should be "good".
But there are shortcomings of this method, the main of which I think
is as follows:
Just the raw number of links does not tell one much about a page.
Basically
the "goodness" we get using such ranks is the popularity of
this page among different
people, who actually can have completely opposite opinions
about the value of the
page or anything else regarding this page.
(It is certainly true that Google
ranks provide additional information about a page
that is not extractable from the page itself
and hence are a great leap in the development of search engines.)
Therefore, with Google you find very popular sites with high approval by the
public, together with the sites highly disrespected by other popular sites,
together with very debatable sites about which *many* people have different
opinions.
And there is no way to tell the difference among these
different possible classes
until you actually go to each site, look at it, and decide on its value
for yourself.
The reason why Google works more or less is that a great deal of the Web
*is* build in the way enabling it to work:
People generally put links to
other sites if they think that the other site is good in some sense.
(This
it basically the same kind of reason why regular search engines
work more or less:
People generally try to give honest and accurate evaluation of their pages.)
But the quality of the search results we can get and the kind of searches
we can make now using Google, clearly indicates that
the current principles behind Google
are not completely adequate for the
Internet searching to be as useful as it can possibly be.
Another point is that an organization with sufficiently high resources
can again easily increase its Google rank just by paying many high
ranked sites
for making a link to the organization's Web site.
Similarly, to get high Google rank a site must only be very popular.
Which in general does not have to imply that the site is a useful source of
any particular sort of information.
Another point is that for example the front page of any big site
automatically gets high Google rank just because many of the pages of this site
link back to it.
Another important feature of Google is that the text of the link
to a given page contributes greatly to the keyword/title description
of the page.
This means that if a page has title "Useful news", but majority of
pages that have links
to it have the link text of "Useless news",
then this page will be more likely found when
searching for "useless news" rather than for "useful news".
Hence, to some extent, what a page is,
is determined by the people linking
to this page rather than by the author of the page itself.
This is another great feature of Google compared to the other
search engines.
But we need a more uniform
way to express opinions about Web pages (not just the link text),
more accurate ways to aggregate
these opinions (not just counting the number of times
a particular phrase is used as the link text), and,
much more importantly, more widespread
use of the ability to express opinions
(I guess not many people are aware that Google uses link text this way).
The two main differences of Clever from Google are that
Such human expert categorization is supposed to be (and frequently is)
better than just traditional searching based on the keywords occurring
on the Web pages. Hence, the popularity of such directories
and their adoption by many major search engines.
(See also http://searchenginewatch.com/sereport/9805-directory.html
for more information on categorizing Web directories.)
The distinctive feature of the Open Directory
(http://dmoz.org/)
is that the editors that contribute to the directory
are not a small hired staff like in the case of Yahoo!
The editors of the Open Directory are
a quickly growing army of volunteers from all over the Globe.
(See http://dmoz.org/about.html
on how to become an editor.)
The Open Directory Project was launched in June 1998;
in February 1999 it had over 7,200 editors and categorizes over 344,000 sites
into over 45,500 categories;
now, July 1999, it has over 12,800 editors and categorizes over 690,800 sites
into over 103,300 categories;
just watch the main page
(http://dmoz.org/)
for the rate of their growth.
See http://dmoz.org/Computers/Internet/WWW/Directories/Open_Directory_Project/
for the press coverage and more information about the Open Directory.
Though, human-maintained directories, especially the ones organized like Yahoo!,
have the following obvious shortcomings:
The Open Directory, having much better ideas behind it than for example Yahoo!,
still has a lot of drawbacks
(all of them can be overcome using the methods proposed
later in this document):
The inability to handle different opinions leads to instability
as more and more new editors join
(confrontation, constant changes in the directory,
and appeals to the moderators.)
Though, the directory database of the Open Directory
is currently available for free to other
companies and people to use under certain liberal license,
I do not think this makes the Open Directory more open with respect
to dealing with the editors creating it.
You might also look at
this part
of the news and history page of this site
for information about discussion of the Open GRiD project
on the Open Directory's forum.
This method definitely reflects *something* about the popularity and relevance
of a particular search result.
But, clearly, the fact that a person has clicked on a search result does not
tell much about whether the person has been actually satisfied
with that search result.
A much better reflection of relevance
(which is currently much harder to automatically gather in practice though)
would be for example the number of people who have bookmarked
a particular search result after visiting it.
The proposed solution is based on a small but crucial extension
of HTML standard
(together with corresponding modifications to a search engine similar
to Google).
The modification of HTML standard is that we add two optional
fields to a hyperlink:
In this case a link from a site www.newsexperts.foo
of the following form
would mean that the authors of www.newsexperts.foo consider the information
on www.somenews.foo regarding computer news to be ranked having
goodness/value/quality of 80% out of 100%.
A link of the form
means that the authors of www.newsexperts.foo
consider international news
on www.somenews.foo to have rank of 10%.
And a link of the form
means that www.newsexperts.foo considers business news on
www.somenews.foo to have rank of -30%.
That is, they think that trusting or using those
news might harm to some extent;
that is www.newsexperts.foo do not
recommend one to rely on business news from www.somenews.foo with
strongness of the recommendation being 30%.
Links without keywords and values would mean that there is no statement
about the contents of the referred page made by the referring site.
Links with these new fields will be called "voting links" or
"opinion links" further in this document.
You can say that these all are very subjective opinions of
www.newsexperts.foo about www.somenews.foo.
That is absolutely true. But when one computes
for example "computer news" rank of www.somenews.foo based on *all*
such opinion-loaded
links from all sites worldwide taking into account "computer expert",
"computer news expert", etc. ranks of all these sites,
one will get the very precise opinion about the quality of computer news on
www.somenews.foo as expressed by *everybody* on the Web who has such voting
links and whose vote either directly or indirectly has influenced the computation
of this particular www.somenews.foo rank.
This rank computation scheme again has a recursive (cyclic)
nature very similar
to the one used in Google.
But one difference is that for each page there might be many ranks
reflecting its value in different categories.
I do not claim that some particular rank computation
scheme is *the* best.
The point is that once we have all this voting data, one
can implement many different strategies to compute ranks reflecting
many different things.
As I see it, with the Open GRiD Project being widely implemented
anybody can do searches like the following examples:
The important thing to note is that,
the results one gets is the best available reflection of the opinion
expressed by everybody who was willing to express it and could physically
do so (using a particular method of a particular search engine to
aggregate the opinions).
Another point is that everyone is free to make any kind of search
and interpret the search results the way he/she wants:
Note also that for example just the fact that an expert expresses
a positive or negative opinion about a Web site with high or low rank
will not influence the rank of that expert directly and immediately.
But the rank of that expert will eventually change if his/her act of
expressing this opinion
influences the opinions of others about that expert.
As one can see the proposed ranking principle is just an extension and
globalization of the way our society determines the value of anything for
any particular purpose.
For example, one can not get high position in any field (especially
in science) until he/she
The positive and negative voting links allow to express and take into account
the precise opinions of people in a much more accurate way that just
the link text.
The Open GRiD Project being widely implemented should be
a very stable system against biased opinions
(see below for more arguments about this).
The proposed voting links when used Web-wide with an appropriate search engine
*are* a perfect categorizing Web directory such that
As mentioned on the Search Engine Watch site
(http://searchenginewatch.com/sereport/9808-clicks.html),
researchers of the IBM's Clever project also see a possibility
of the Clever project methods being applied to creation of human-less
categorizing directories.
The main innovations of the proposed solution
over previous work on this subject are
The important fact is that as long as one does not have
*good* means to rank, order, and weight the *authors* of opinions
any system serving user generated content
other than just ranks (e.g. comments, classifications, etc.)
is going to become more and more useless as the number of contributors grows:
The users will be buried under the information for which
it is not clear which parts of it can be trusted.
And a ranking system will provide just primitive public poll results.
Here is a possible scenario how one can create votes (that is express his/her
opinion) very easily (almost as easy as bookmark creation):
There are no real technical problems here:
Check regularly for news
and updates
or simply subscribe for
the news mailing list.
The Problem
The problem (which I think is the main problem) is as follows:
It is improperly difficult to efficiently find all the relevant information
one can look for that is available somewhere on the Internet.
And one never knows how good is what he/she has found
and what he/she has possibly missed.
That is, the currently available search engines and directories are not adequate
for the ever-growing complexity and size of the Web.
Towards a Solution
In this section we discuss several existing approaches on how to radically
improve the quality of search results.
(A similar review-style analysis of most of the approaches discussed below
is given on the Search Engine Watch site:
see http://searchenginewatch.com/sereport/9808-clicks.html.)
Google Linking Ranks
Google Link Text Consideration
IBM's "Clever" Project
A very similar to the Google search engine is the IBM's
project called Clever
(http://www.almaden.ibm.com/cs/k53/clever.html).
This project also uses both link ranks
and link text in a fashion similar to Google.
Hence, all the reasoning above about Google can be applied
almost without changes to the Clever project.
This should make Clever's search results more accurate
at the expense of very high resources/time
required for interactive searches.
Human-Maintained Directories
Categorizing Web directories like the Open Directory -former NewHoo-
(http://dmoz.org/)
and Yahoo!
(http://www.yahoo.com/)
are trying to solve the Web searching problem from another side:
They provide listings of many sites categorized
by many human experts.
Hence, only directories build by *very many* volunteer editors
can create good categorizations for the whole Internet.
For example, already now (February 1999) Yahoo! is forced to constantly
decrease the percentage of sites that get categorized out of all
sites wishing to be categorized.
That is, they are forced to keep their directory more and more incomplete.
See http://searchenginewatch.com/webmasters/yahoo/delays.html
for more details.
Thus, only directories build by appropriately growing number of
volunteer editors will eventually survive as the Web grows.
In the case of the Open Directory an attempt to do more ranking
will just make the directory very chaotic and
incur more "wars" between the contributors to the directory
(called editors in terms of ODP),
because currently (February 1999)
the Open Directory Project does not have means to support
ranking nicely.
This is valid even for the Open Directory how it is organized now:
Note that the comments below are based on my short experience of being an editor
of the Open Directory. I think they are correct conclusions from
what I have seen,
but maybe sometimes too gloomy conclusions from things which are just temporary.
The way how editors can expand their editable area
and the way to build a good hierarchy of editors
(that is to rank the editors)
seems to be obscure or not yet determined
(now it is done solely by the top moderators).
It can not handle "unlimited number" of editors
because of its centralized implementation and
centralized moderation.
According to Rich Skrenta, the main initiator and moderator
of the directory
(http://dmoz.org/profiles/skrenta.html),
currently they are rejecting 66% of the applications
to become a new editor,
though maybe those are mainly the applications for too broad categories
which can not be accepted just because of the inability to handle nicely
opposite opinions at high levels of the directory
(See http://dmoz.org/forum/threaddisplay.cgi?t=Forum1/HTML/000219.html,
but you have to be an editor to access that document.)
The way sites are listed in a particular category is
determined by consensus of all the editors who have
the right to modify this category.
That is, if there are different opinions the editors have to
somehow find a compromise and present *only one*
opinion to the users.
Although an editor can set up his/her own bookmark directory
with his/her own opinions,
those bookmarks are not much more a part of the directory than
a regular home page.
Hence, the choice might be between supporting diversity and
stagnation because of the inability to handle more editors than
a certain number.
Link Click Counting by Direct Hit
Direct Hit
(http://www.directhit.com/)
is using the number of times people have clicked on a particular search
result to reflect the relevance of the search result
for a particular search query.
The Proposed Solution
See also section
"Possible Extensions and Topics to Discuss and Settle On"
later in this document for some other interesting
extension that can be made.
<A href="http://www.somenews.foo/" cat="/News/Computers"
rank="80%">Good computer news</A>
<A href="http://www.somenews.foo/" cat="News/World"
rank="10%">Some world news</A>
<A href="http://www.somenews.foo/" cat="/News/Business"
rank="-30%">Misinterpreted business news</A>
The point is that you determine what you want to find and how to use the
results, the Open GRiD Project implemented will simply
give everyone the ability to make those searches and
get those results easily.
How the Proposed Solution is Related to What We Know
Society
Google and Clever Search Principles
The proposed solution gives categories to Google-like ranks
to make quality/relevance ranks in a particular category out of
just general popularity ranks.
Human-Maintained Directories
The Open GRiD Project allows *everybody* (without anyone's permission)
to become an expert/editor ranking and categorizing Web sites in a way
that produces consistent and stable aggregated opinions/ranks/categories.
The number of these people is growing together with the Web
and is the only one that can be enough/adequate to produce very
good categorization and ranks in very many fields.
Link Click Counting
Clearly the proposed voting links
(which can be created with the same ease as bookmarks: see the next section)
provide a much better reflection of relevance/quality/value/etc. than
just the number of people who click on a particular search result.
The Main Innovations of the Proposed Solution
Hence
and these qualities increase as the number of participants grows!
Examples include newsgroups, Crit comments
(http://crit.org/),
SlashDot comments
(http://slashdot.org/),
Deja.com comments and ratings
(http://www.deja.com/),
Alexa reviews
(http://www.alexa.com/),
etc.
Hence
Hence
Easy Vote Creation
It is an even less problem to create a special
OpenGRiD-aware client proxy
(see the
Software Download Page to get such a proxy).
 |
Quick link to "What *You* Can Do Now" section. |
Here is an incomplete list of responses I have gotten so far. (I will be updating it from time to time.)
See http://dmoz.org/forum/threaddisplay.cgi?t=Forum3/HTML/000031.html for a discussion about the Open GRiD project on the Open Directory's forum. (But you have to be an ODP editor to access that discussion; here is a local readable for all copy, which is most probably not up to date.)
You should also look at this part of the news and history page of this site for more information about this discussion.
Note that from Dec. 22, 1998 to Feb. 12, 1999 this document was tentatively titled "The Christmas Document" according to the time it was first posted.
This is a response from Sergey Brin, president of Google Inc. (http://www.google.com/), for an announcement of this document (posted with his consent):
Subject: Re: News: The Christmas Document
Date: Thu, 24 Dec 1998 15:12:39 -0800
From: "Sergey Brin" <xxx@xxx.xxx>
To: "Maxim L. Lifantsev" <maxim@cs.sunysb.edu>
Your ideas sound remarkably close to what we discussed in research meetings at
Stanford. I think something along these lines is certainly a good idea. The
main obstacle I see is the chicken and the egg problem. It is hard to get
people to use this standard until it is already being used.
Good luck,
--Sergey
Sergey Brin
President,
Google.com
The LinuxMall (http://www.linuxmall.com/) has published a link to this document. This is their announcement:
The Christmas Document
Ed: An interesting whitepaper about Internet search engines and
how they work versus how they should work. The document
requests an extension to HTML in the process. Judge for yourself.
Thanks to Maxim Lifantsev for this link. (December 1998)
You might also wish to look at the mailing lists archives.
 |
Back to the beginning of the document. |
You might wish to look at these sections of the Frequently Asked Questions:
That is, they have a general computation scheme for the whole Web, but no categories and no evaluation votes. (Though taking link text into account which they already have is a step towards categories and evaluation votes.)
They categorize (and rank a little) a particular subset of the Web using the efforts of a very limited number of experts chosen by the organization maintaining the directory.
They categorize (and rank a little) a growing subset of the Web using the efforts of very many self-elected editors which are selected by the moderators of the directory using a not quite clear (but liberal) principles. (See this part of the document for more details.)
That is, they cover one particular category: operating systems; there are only two possible types of evaluation: rules or sucks; all votes have equal weight because there is no way for them to distinguish between experts and non-experts, for example.
This is again just equally weighted public opinion on particular subjects with only one type of vote: "I support".
That is, this method ranks some pretty large fields (CS or physics research publications) using positive votes weighted according to reputation of the place where a citing paper has been published.
Many of the following items are already discussed in detail and/or their solutions are proposed in The Open GRiD Project Architecture Proposal (which is a rather technical document):
This way the search engine for the Open GRiD Project can collect all these descriptions and show the one with the highest rank along with the link to the other descriptions. Since people ranking a particular site can easily find each other, they can coordinate in order to each provide the same description a certain group of them agrees on.
What is the exact definition of "a site" whose ranks we are computing? Is it just one Web page? or a whole site? How to determine the borders of "a site"?
How does the rank of a Web page in one category influence the weight of opinions expressed on the page in other categories? What is the degree a two given categories are related?
Actually, I think once there is at least one such more or less plausible method, the system will be able to improve itself by means of people expressing opinions about what method of aggregating opinions is the best.
Also when XML and RDF become widely supported by browsers one will be able to formulate the needed extensions within the standard.
That is, the assumption is that many people given the ability to vote on any subject they wish (with an easy way for everyone to find out about the results of the votings) will do so.
This assumption is confirmed 100% by the number of editors of the Open Directory (http://dmoz.org/) and especially the rate of the growth of that number, even though the Open Directory has a certain set of limitations including an estimated 66% disapproval fraction for the applications to become an editor.
Another good evidence that people will create enough of such voting links
is this:
I guess almost everyone who surfed the Web a lot has
(a lot of) bookmarks.
I guess very many of such people are willing to (and do) make their bookmarks
publicly available to help (advise) other and allow them to use
one's efforts spent surfing and creating bookmarks.
The problem is that currently there is no uniform way to
help many new surfers utilize this knowledge accumulated in
bookmarks (i.e. opinions) of many people.
But with the Open GRiD Project implemented everyone will be able to easily make his
opinions (that is basically rank-loaded bookmarks)
both available for others to use and more automatically categorized for
the person him/herself.
But it is impossible to implement this solution instantly: many people must create sufficient number of voting links on their Web pages for the proposed method of searching to be useful. Therefore, I think there is no danger of fast sharp radical changes in the society possible caused by implementation of the project.
This is very similar to the free software communities like the Linux operating system development community (http://www.linux.org/):
In an environment where people's reputation is determined by the evaluation of all the others members of the community, people are very careful with their reputation and their statements (and in general with any product/idea they make widely available) because every time a person makes a statement it is his/her reputation what is at stake.
As a result such communities produce very good (for the community) solutions/products/decisions because a solution/product/decision will be rendered good (and hence will be widely adopted and will promote the reputation of its creators) only if the weighted majority of the community agrees on that.
This also explains why Linux has been a system more for a computer expert than for an ordinary user: the Linux development and ranking community consisted so far mostly of computer experts.
See these links:
http://tuxedo.org/~esr/writings/cathedral-bazaar/
http://tuxedo.org/~esr/writings/homesteading/
for more information on how Linux community works.
On the other hand, several people have suggested to me that there is a danger of this voting system to be biased.
One possibility is a company spending a large amount of money to create many sites with favorable ratings for each other and for that company to construct a more or less closed self promoting set of sites or an advertisement network trying to increase ranks of firms that are paying for favorable votes.
Some reasons (in addition to the ones discussed above) why that should not work are as follows:
First, this problem existed in our civilization ever since. And the answer is that, if the idea applies to a particular field, then its author must make somebody known in the field see the importance of the idea and help make it discussed and then accepted if the idea is worth it.
The good point about the Open GRiD Project is that it will make it possible to easily search for unaccepted new ideas. So, if somebody thinks that it is his/her duty to help out new ideas get accepted, he/she can systematically look for such ideas and help the authors of ideas to get in touch with appropriate reputable experts in the field who can support the idea (or archive these ideas for future consideration). Similarly the author of a new idea can easily look up appropriate experts in the field him/herself.
Also this paragraph might provide a (partial) solution.
This document will also be a live example of whether it is currently
possible
for an idea with a potential for worldwide importance to get widely accepted
starting with almost zero reputation and famousness of the author:
I am a 25 year old Ph. D. student in Computer Science at State University
of New York at Stony Brook
(http://www.cs.sunysb.edu/)
specializing in a pretty narrow field of
theoretical
computer science.
My one year old home page
(http://www.ecsl.cs.sunysb.edu/~maxim/)
got about 800 hits during 1998 (that is about 2.5 per day)
mostly by students from Stony Brook.
This is not surprising because my home page almost did not have any
useful content
except my papers which might be interesting for a very small number of
people.
Also I have never before submitted my home page to be indexed by a search
engine.
Either one expresses his/her opinion anonymously as just an ordinary person (as we all do when we vote in elections); or one makes his/her statement public and uses his/her reputation as an knowledgeable person in the area to increase the significance of the opinion.
This is no worse that the current state of the things in the world which are exactly as described. The difference is that with the Open GRiD Project implemented, *everybody* will be able to voice his/her opinion (if he/she wants to) either anonymously as an ordinary person or publicly as an expert on *any* question. Another difference is that *anybody* will be able to find out what is the opinion of anybody on any question which he/she made public.
The condition of making expert opinions public is actually
a required precondition for such system to be stable and valuable.
If anyone can use his reputation to influence something, so that
nobody knows in which way he/she has influenced it, then
it is not possible to trust all these ranks, because they might be
biased: In this case an expert can lie about anything
without any danger to affect his/her reputation.
Actually, it might be possible in principle to implement
anonymous electronic
voting both as an ordinary person and as expert enforcing the rule that
everybody can generate only one vote as an expert and one vote as
an ordinary person on any question.
The idea is to use public key cryptography
(see for example
http://www.pgpi.com/)
to identify people and associate
them with their reputation as experts.
The second component is to use some free (open source) software
(http://www.opensource.org/)
to process the votes and ensure anonymity and the "only two votes" property.
The main feature of free (open source) software needed here is the
availability of the source
code for such counting and vote collecting software.
This availability will guarantee that
the software really does all the best to ensure anonymity of the voting
and does not cheat.
But even with the source code available everybody
has to trust the actual people running the vote aggregating software,
because since the votes are not available for everyone,
it is not possible to check that the results generated by
the vote aggregating software are correct.
To increase the credibility of results is is possible to use many
such systems run by different people, but still the force needed to influence
any small number of such systems is quite different from the one
needed to influence the whole Internet.
Hence, such anonymous voting systems should be used only
if there is no other choice like in the scenario described
in the next section. And the results of such votings must be
always considered to have much less credibility than then ones
produced by openly expressed opinions.
Many people of a democratic to some degree country openly express their political opinions on the Internet. Suddenly the country becomes a dictatorship. It will be trivial for the dictator to find all his/her opponents using the Internet with the Open GRiD Project realized.
If you think that this is a too farfetched scenario, you are welcome to argue that and propose a more realistic scenarios. Another point is that it is only good if this scenario is farfetched.
The good thing is that nobody has to express his/her opinion if he/she
does not want to.
Another point is that it might be possible to
use a trusted anonymous vote collection system
(as described in the previous section)
to collect opinions anonymously in such sensitive cases.
Though any such system can be broken into in principle.
Looks like such an anonymous opinion
collecting system can have two architectures:
Let me know if you came up with other major misuses so that I can augment this section.
If you think about it, it is clear that if the idea of the Open GRiD Project is correct, it is inevitable that somebody comes up with it at some point in the evolution of Internet; and it is inevitable that this idea is widely disseminated and implemented once made sufficiently known.
The possible sub-scenarios if the idea is correct are:
It is natural to suppose that many big firms or governments might not like this idea of giving so much power of making decisions and influencing the civilization to the people. But no single firm or government will be able to stop this, because by now the Internet is global, that is, it is much bigger than any firm or country. And the Internet is not and hopefully can not be completely censored or controlled by any firm or government. Also I think such firms and governments will not be able to unite their efforts because they are competing against each other.
Hence, it should not be a problem to support both the people actively working on the Open GRiD Project and provide them with the necessary equipment once the project gets sufficiently popular. Actually, the "banner" way to get the resources might well be not even necessary.
|
|
Quick link to "What *You* Can Do Now" section. |
With the current size of the Internet and the Open GRiD Project widely implemented the chances of a lie to be discovered in a fairly short time and established to be such if it really is a lie (that is according to the overall interested people's opinion) will be pretty high I think. This will make everybody think very carefully before making a false public statement about anything.
One widely understandable example might be Microsoft (http://www.microsoft.com/): The Open GRiD Project realized will do *one* of the following:
It is also true that the Open GRiD Project realized will be a public extension of commodity protocols (as opposed to extension pushed by a corporation) and will be a good answer of the public to the idea expressed in the Microsoft's leaked memo named "Halloween II" (http://www.opensource.org/halloween2.html) regarding a strategy to try to own and control commodity protocols and hence the Internet.
If a company thinks that the ranking methods of the search engines determining that value of their products for the society are wrong, they are free to ignore them. But if the company is wrong about this, it will lose to a competitor.
In general, I think that the Open GRiD Project implemented worldwide will be a *very* stable and *self-correcting* and *self-improving* system.
Copyright (C) 1998-1999 Maxim L. Lifantsev
This refers to the copyright on the ideas of the Open GRiD Project expressed in this document.
The initial solution has been generated by the author, Maxim L. Lifantsev, on December 16-18, 1998.
Since the idea of the Open GRiD Project is I think very simple when you know it, I am not certain that I am the first one to come up with exactly this idea or with a very similar one, but I created this idea independently based on what is mentioned in the acknowledgements section.
The license guiding the usage of this document describing the Open GRiD Project is almost identical to the GNU General Public License (GNU GPL) (see http://www.gnu.org/copyleft/gpl.html).
Main implications of that are as follows (the differences form GNU GPL are also indicated below):
You are free to copy this document in any form
provided you copy it
verbatim including this notice and the date the copy has been made.
For example, anybody is free to just publish this whole document
in any journal without asking (or a reimbursement
to) the author of the document, provided the other requirements of
the license are fulfilled.
You are free to elaborate on this idea and implement any software, hardware, etc. based on the idea or write explanations, additions, translations, etc. to this document provided:
If you would like to see the project implemented,
it is important that you do your (small) part of the job
to make it happen.
The point is that the Open GRiD Project can not be realized by
one person or by a small group of people without some support
of many people.
Active participation of many different people in adoption and
use of the project is a necessary precondition for it to come true.
There is no real reason why you can not (or should not) be among
the people helping the project to become a reality!
The easiest thing to do is to vote in favor of the ideas of the Open GRiD Project to let other readers know about your opinion. Just click on the the button below:
The current results are as follows:
|
votes supporting the Open GRiD Project |
|
votes against the project |
|
votes totally |
Certainly these voting results are not very useful because (1) this voting does not change much; (2) many people just do not vote at all; (3) you do not know who voted and why he/she voted in a particular way.
The next section lists more interesting and helpful ways
one can support the project.
The simplest way is to submit (a link to the page with) your
opinions about the project to me
or to the appropriate mailing list
by e-mail;
or simply by visiting this document following a link
from the page that contains your opinions
using a browser that provides the referring page URL.
Most probably it will appear on
this page
soon.
(Later on you will need to submit the link
to a repository of pages related to the project
once such repository is established.)
The good thing about these methods
is that your page will be instantly accessible once included.
The price for this is that any such repository will be inevitably moderated
by its maintainer to some extent.
Here are four other ways to get some reflection of opinions of others:
The first way that is already implemented, but which is the least accurate,
is to look at the access counters
for this page.
(Some third-party counters provide
much more statistics than just the raw number of hits.)
The second way is to find this site using the Google search engine
(http://www.google.com/search?q=open+grid)
and look at the site's rank according to Google
(see the discussion on Google
above to find out how much this rank will mean).
The third way is to look at the
vote results
regarding the project.
The fourth and the most accurate way will be to use the search
engine of the Open GRiD Project itself to get
the popularity/value/etc. statistics
about the project as it gets implemented more and more widely.
I guess I will have to augment and make more detailed
this section as the popularity of the Open GRiD Project grows
and more people contribute to it.
Here are some of these "post-creation" acknowledgments:
Anybody is welcome to send me
(maxim@cs.sunysb.edu)
an e-mail with information
you think I must know or that must be incorporated into this site (points
than need to be changed,
missing points, wrong points, etc.).
If you have some message appropriate for one of
the mailing
lists dedicated to this project, submit it there.
BUT, please be thoughtful: do not abuse my e-mail address
or the mailing lists.
Concrete Things to be Done
These are the main areas in which some help is currently needed:
Suggested places are news sites and news groups
whose readers are familiar with the free (open source)
software movement (e.g. Linux related news sites).
Here is the first such non-technical text (very alpha):
http://www.sinc.sunysb.edu/Stu/achuvaki/Xmas/
What has been Done
How to Make Your Opinion About the Open GRiD Project Heard
Means to Find Out About Current Degree of Acceptance of the Open GRiD Project
I guess you have already looked at section
"What People have Said About the Open GRiD Project"
above.
You might also look at the project's
news and history page
and mailing lists archives
for the news coverage, discussion, and other opinions about the project.
Acknowledgments
I would like to mention people who have somehow
contributed to creation of this document:
And all the people helping to distribute such
information.
On Sending E-mails to the Author of This Document
Table of Contents
|
|
Back to top of this page. |
|
|
Back to the main page of the Open GRiD project. |
 |
To the author's home page. |
 |
The first version (0.1) of this document
was first posted on Dec. 22, 1998
This is version 1.0 possibly with some modifications that do not yet qualify for a new version number. It was first posted on Feb. 13, 1999 and last updated on Dec. 17, 2000 at 04:29 PM EST by Maxim Lifantsev Here is the revisions and changes information. Today's date is Dec. 05, 2008 |
| For disclosures about the properties of the pages on this site look here. | |
 |
Comments, Suggestions? |