All people belong to communities that express collective principles using documents. Existing solutions to change or create documents, such as representatives, petitions, and town halls, have problems. Certain people with more power or resources are unduly influential. And the ability to scale to larger, diverse, multi-lingual populations is rarely present. The documents we live by, therefore, cannot be created or changed without great struggle, even when the need arises. Wirite solves these issues by giving control to people. Users create, vote on, and express issues with edits. This information is then used to place the ideal number of non-conflicting edits into new document versions. Crowd-sourced translation of edits allow the document to exist in multiple languages throughout the entire process. And document creation and community support occur simultaneously, reducing the ability for unbalanced influence from outside sources.
Night 7k Call Girls Noida New Ashok Nagar Escorts Call Me: 8448380779
The Document Project
1. The Document Project
www.documentproject.com
Andrew Inglis
1 Introduction
When people form groups in order to live with and care for one another, stand collectively behind a cause,
and work towards a common goal that they believe in, the results can be powerful and revolutionary.
This has been a theme throughout history. Some of the groups that we see in the world are: friends,
families, organizations, teams, communities, religions, unions, governments, and countries. Within such
diverse and dierently sized groups there are similarly dierent time scales with which they exist - from
hours to generations - dierent connections - ranging from loose aliation to strong interconnectedness, and
dierent enrollments, as people's needs and desires change, and as they grow. In essence, the groups that
people form are as richly varied and dynamic as life itself.
For groups larger than friends and families, one of the most common methods of expressing their prin-
ciples, collective abilities, and proposed actions is by creating a document. Examples of such documents
are: a government's constitution and laws, mission statements of companies, and petitions. Changes to the
documents usually don't occur without great struggle from members of the group, even when the need arises.
This is not a necessarily bad attribute of groups and the documents that they uphold. If we are continually
changing the methods that we live by, then we may never have enough stability to progress as a civilization.
There is a certain level of consistency that is desired and needed. Sometimes, however, the forces to keep
this consistency can overwhelm even the most logical and needed attempts to change systems or sets of rules.
Another point to make about these documents is they are rarely created by the individuals within
the group that support them. This lack of community input is inevitable when documents are followed
for generations, but is also something that is thought to be necessary when dealing with large groups of
people (for example, more than 500 people). It is hard to imagine having millions of people contribute
in a meaningful way to the same document. Instead, the document is created by a small group of leaders,
activists, or representatives in a manner they feel will garner the support of the larger group. Although there
are leaders in our society attempting to be the voice for large amounts of people, there are inherent problems
that single individuals face when trying to represent very large groups of dynamic, changing people.
The question I am trying to answer is: What is the best tool available, within current limitations of
technology and communication, to allow a large number of people with disparate ideas and backgrounds to
coherently and sensibly interact with one another on a large scale when the need arises? And how is the
1
3. t of this interaction maximized by making the contributions from each individual as self empowering
as possible?
The purpose of the Document Project (DP) is to search for an answers to these questions. Its goal is
to allow the editing by and contributions of people to a single document at a scale that is unsupportable
by traditional document sharing platforms (such as Google Documents and wiki systems). The platform
attempts to allow 2 people to 7 billion people edit, contribute, and feel a sense of ownership in a single
document that could be as short as a paragraph or as long as a novel, thereby allowing more people to be
invested in a collaborative treatise of shared interests and action, with the real potential of the tool showing
at the level of 20 people or more contributing. This is something that no other collaboration software allows
at this point.
2 Background
We are most familiar with one person creating a document that expresses their ideas, thoughts, and proposed
actions to the world (Figure 1).
Figure 1: Single person creating and editing a document
The tools that allow a person to do this are numerous: a piece of paper, a typewriter, a word processing
editor such as Microsoft Oce, an Internet publishing tool such as a blog, an email account. Since a single
person has complete control over the creation of the document, she can write the document in a way that she
sees as valuable as possible for herself and the people that she wants to share it with. Many of the editing
tools created have focused on one person creating content and sharing the content with the world.
The same goes for a small group of people of less than 4 or 5 (Figure 2).
Figure 2: Four people creating and editing a document
In this case, it is also relatively straightforward for a group of people to use the same types of systems that
the individual person use (like a word processor such as Microsoft Word) to create a document that mutually
maximizes their satisfaction. This is because it is possible for the entire group of 4 or 5 people to get together
2
4. and talk to each other at length during the editing process, and settle to a consensus about what to write.
It is still possible then for satisfaction to be high in a group that gets together and creates a document that
expresses themselves as a collective.
Figure 3: Twenty or so people creating and editing a document
For a larger group of
5. ve to twenty people, things can get a little more dicult and complicated. It is tricky
to get everyone together at the same time, and when that is possible, it is dicult to come to a consensus
with such a large group. One easy way is for a single person to hear the suggestions of the larger group,
and incorporate those changes into the document. Conversations can happen about what to change, but
even this can get tricky when the number of proposed edits become overwhelming to discuss. There are
several collaborative document editing programs which attempt to ease the discussion of such sized groups:
the programs store proposed changes from members of the group and facilitate discussion about the changes
in an orderly way, so that the controller of the document can make a
6. nal decision about how to make the
changes. Such collaborative software programs exist as separate entities, and are also provided as add-ons
to the more traditional single person editing programs mentioned above1.
Figure 4: A large group of 100, 1000, 10000, or more people.
The situation gets even more complicated when 100, 1000, 10000 or more people want to contribute to
a single document (Figure 4). Here are a few methods that are used to handle this endeavor:
Wiki Systems
Wiki systems, such as Wikipedia, allow the world's Internet users to edit documents together. Anyone
who wants to change a document is free to do so, anyone else can decide that it is not a good edit and
remove it, and a conversation occurs between the two people in a discussion site along with the rest
1See http://en.wikipedia.org/wiki/Collaborative real-time editor/ for a list of software available
3
7. of the community interested in that topic, and this cycle continually repeats itself and the documents
continually change2. This system is currently used to create encyclopedias of knowledge: a set of
articles, where each article treats a speci
8. c topic independently and distinctly.
Petitions
Petitions are documents that expresses a group's principles and collective abilities. The documents are
usually signed by the numerous individuals that back them. The life-cycle of a petition is usually as
follows: an individual or small group of 10 or so people create a document that they feel will garner the
support of a much larger group of people. They then oer the petition to the larger group to pledge
support for the ideas expressed in the document. The goal of almost every petition is to maximize
people power by having the strongest message possible and having the largest number of people
signing it.
Representative Democracy
Systems of government collectively state the intentions of a large group of people by having represen-
tatives selected periodically by smaller subsets of people to represent them for a given period of time.
The representative's job is then to come to agreement with one another and create rules or a document
that the entire community then abides by. Anyone who wants to change a law can petition represen-
tatives with their viewpoint, vote dierent representatives in, or become a representative themselves.
We see this type of system in small organizations, to local governments, to national governments, to
the United Nations - it is one of the most widespread democratic systems that we have in our world
today.
When discussing wiki systems, petitions, and representative democracies, it should be stressed that the
point is not to make arguments for replacing them. These systems are positive and integral parts of our
culture. Rather, critiques allow us to explore new options for collaboration in our communities, and the
possibility of
9. lling needs that older systems are not providing.
In beginning to evaluate the pros and cons of systems that can help bring together large groups of people
to support a cause, it is worthwhile to think of the following attributes of such systems:
Does the system have leaders that have more control over the document than the people
that will support the document?
We are very used to people taking the lead and attempting to be the voice for large amounts of people.
Lots of good can come of this, but it is worth asking How much potential do leaders have to overly
express their own ideas in the
11. ts of leadership, especially when it can distance the people of the community from understanding
issues that directly aect them and decisions made on their behalf. One goal of any healthy society is
to empower people to be knowledgeable and control their own lives and destinies. When evaluating a
system that attempts to express the collective ideas of a group, leaders in
uence should therefore be
evaluated.
2Wikipedia advocates a so-called BOLD, revert, discuss cycle - meaning be BOLD with the edits you make,
wait for someone to remove (revert) your edits if there is a disagreement, then have a discussion about. see
http://en.wikipedia.org/wiki/Wikipedia:BOLD, revert, discuss cycle/
4
12. Does the system let certain individuals in
uence the document more than others if they
want to?
When people get together to do things, it is common for those with more time or intention to put more
eort in. Just like having leaders, this can be a good thing, since passionate people usually do a good
job handling things they are passionate about. This type of action is so common in our societies that
we assume it as given. However, when creating a document that expresses the collective ideas of a
group, it is not as democratic to allow very uneven contributions. There are many reasons why people
don't contribute as much as others even if they are just as interested in sharing their voice: they could
be too busy with other things in their life; they may not be informed as much as they would like, or
they may be unsure of their abilities. When evaluating a system that attempts to express the collective
ideas of a group, it is worthwhile to see how accommodating the system is to a wide variety of people's
abilities and time to contribute.
Does the system handle the many voices and contributions from it's users, even as the
group gets larger and larger?
Systems that work just as good when there are 100, 100000, a million, or a billion people using them are
called scalable. Scalability is a good thing since it is unfortunate when a system that is working well at
one size, say with 1000 people involved, stops working when 10000 people get involved, especially when
the goal of the system is to include more and more people. When evaluating systems that attempt to
express the collective ideas of a group, it is worthwhile to imagine how well it works as more and more
people get involved.
Does the system allow for ideas that are in the majority to suppress ideas that only a
few people have?
The so-called tyranny of the majority is a worry that in any system, decisions made by a majority under
that system would place that majority's interests so far above a dissenting individual's interest that the
individual would be actively oppressed3. In addition to creating an oppressive environment for people
with non-majority ideas, this also leads to a narrowing of viewpoint by the community since ideas
from diverse minds are not being expressed. Most importantly, it disallows very dierently minded
people from coming together on a given cause only because one group has more people compared to the
other. Allowing ideas that less than half of the people support to fairly in
uence the document creation
process is a fundamental pillar of building consensus within diverse communities. When evaluating
systems that attempt to express the collective ideas of a group, it is worthwhile to see if such inclusion
is integral to the process.
Does the system allow people that speak dierent languages to edit the same document?
Systems that can create groups that are inviting to all languages are important, since the language
dierence should not get in the way of people standing collectively behind a cause. If the system
is scalable and able to represent non-majority beliefs fairly as mentioned above, then the system
could accommodate unprecedented numbers and diversity of people. With the ability to have people of
dierent cultures and languages edit documents with one another, the groups that de
13. ne the traditional
3See http://en.wikipedia.org/wiki/Tyranny of the majority
5
14. structure of our communities (such as nations and religions) become less of a barrier between people.
When evaluating a system that attempts to express the collective ideas of a group, it is worthwhile to
ask whether it can allow for this freedom across languages.
Figure 5: The case of representation: wiki systems, petitions, and representative democracies
The three tools of wiki systems, petitions, and representative democracies highlighted above have some
similarities concerning the challenges just mentioned. All three systems use a subset of the population to
create the document of interest (Figure 5), whether it be an encyclopedia article, a petition, or a document
of governance. For a wiki system, although anyone in the world can edit a given document, the number of
people that contribute to a given Wikipedia article is around 300 or less, and there are not any articles that
have thousands of active contributors at any given time4. For a petition, it is the group of activists that
create the document in the
15. rst place - less than around 20, and for the representative democracy, it is the
representatives in the governing body, which is less than around 500. These numbers show that while the
amount of people that can be represented by these methods can grow, the number of active creators of the
documents are staying less than 1000 - the amount of contributors is not scaling. This leads to the following
speci
16. c observations of these systems abilities to bring together large groups of people:
Representative Democracy
The more diverse and large a community gets, the more dicult it is for a system to keep the number of
representatives to a constant minimum of around 500 or less and still enable the representatives to work
with each other to come to consensus on important issues. This could be because of the complicated
pressures representatives feel during the process of becoming elected. One of these pressures is that
of smaller groups of people which use resources to in
uence the representatives in a disproportionate
way. Lastly, it is dicult from within a given governance system to reach solidarity with others from
another governance system, even if such overlap would be helpful to both groups.
Petition
A petition can be translated into multiple languages for people to read, and it can be read and signed
by an unlimited number of people. Also, there are two forces that minimize the chance for non-majority
ideas to be suppressed. First, in their attempts to collect the most amount of signatures of support
for the document, the drafters create the most accommodating document, and therefore attempt to
4See Druck et al. Learning to Predict the Quality of Contributions to Wikipedia.
6
17. incorporate non-majority ideas. Secondly, the very nature of a petition is such that whomever decides
not to support and sign the petition can walk away and not support the cause. By doing this, they
gain nothing, but are not forced to support actions they don't agree with, either.
The maximum number of people that can get together in this traditional method and the maximum
power for change that the document can create, however, is usually not attained for several reasons.
First, any document has multiple points, some of which have the potential to make one supporter or
another feel less interested in supporting the document - so-called section cooling. If the document
is written less controversially in order to collect more support, then it lessens the document's ability to
invoke change, even if more people sign, which may not be the case since a watered-down document
is less inspiring to support. Also, after the petition document is created by the leaders, there is a
certain amount of time that is needed to garner support and have people sign the document. During
this time, similar to a representative democracy, smaller groups of people that have disproportionate
resources can unduly in
uence the discussion of whether the larger community should sign the petition
or not. A last pitfall is that even when people sign the document, pledging their support, they are
not invested 100% in the cause, because they had no say in the creation and development of the
document, and could not express their detailed ideas towards the multiple concepts of the document.
With these challenges in mind, the small group of writers must
18. nd a perfect balance by having an
intimate knowledge of all of the supporters for the cause, or have the ability to perform large amounts
of polling research, or, in a shotgun approach, submit numerous documents and hope that one that
catches the minds of a group of supporters. These challenges reduce the power of the petition as a
mechanism for meaningful change in our society.
Wiki
Wiki systems allow for an unlimited number of people to contribute in a meaningful way to factual,
historical, and scienti
19. c documents. Unfortunately, moving from the discussion of these topics to that
of governance and opinion, where unlimited numbers of people are contributing ideas, is not possible
on wiki systems. What does happen on wiki sites is very similiar to representative democracies and
petitions: experts on a given topic are the self-selected representatives for the rest of the community.
Such a system is a natural method for scienti
20. c or historical topics where there are a handful of
experts, but does not work well when dealing with such opinionated topics. The alternative to having
representation, where hundreds of thousands of people voice their opinions by editing an article, is a
very dierent scale from the 300 or so editors that the wiki system currently handles on average for each
document. In essence, wiki system cannot handle the volumes and frequencies of contributions that
would come if as many people that read the documents also contribute to them. In fact, Wikipedia is
the
21. rst to mention that it is not a democracy or a place for original thought or opinions5.
3 The Document Project
The Document Project is an attempt to democratically bring people together to write a document and follow
the fundamental ideas of being controlled by the people that will support the document, having fairness of
input, scaling to any number of people, and allowing diverse groups to contribute.
5See http://en.wikipedia.org/wiki/Wikipedia:What Wikipedia is not/
7
22. Figure 6: Goal of the Document Project
Representative
Democracy
Petition Wiki Document
Project
documents controlled by
followers of the document
5 2 5 7?
fairness of input 5 5 5 7?
scaling of people that can
4 2 4 7?
give meaningful contribu-
tions
inclusion of minority ideas 5 6 5 7?
diverse people contribut-
4 6 8 7?
ing
Comparison of dierent methods of community expression. Scales are qualitative from 0 (poor) - 10
(excellent). The Document Project attempts to maximize these numbers with current technologies as best
as possible (represented by all 7?s).
The Document Project adapts the traditional petition style of community expression so that the
23. nal
document expresses a set of collective beliefs and actions of the people supporting the document in the most
powerful and in
uential manner possible. Enabling the backing of the document to occur at the same time
as the formation of the document is a fundamentally new concept in petition-like documents, and has the
potential to reduce the issue of small resource-rich groups being able to manipulate the larger support group,
since the documents inception, development of support, and pledge of support occur simultaneously. Also,
the method solves the problems of section cooling, signer's investment, and maximizing overall document
in
uence (number of people signing and potency of document) by allowing all
24. nal supporters of the
document to play an integral role in the drafting and critiquing of the document. Lastly, it allows contributors
to express dissatisfaction throughout the process, empowering each individual as much as possible in a
democratic setting.
The ideas in the preceding paragraphs are only possible by creating a document sharing system that
scales to an unlimited number of people. This is challenging becuase such a system must collect the changes
that are made by the individuals contributing to the document, but if the number of active contributors is
in the thousands, the amount to discuss and come to consensus and conclusion on also becomes so large that
8
25. it is dicult for each member to evaluate each change that is made 6. This is the main problem that wiki
systems face when attempting to be a document creation system for large amounts of contributors.
The Document Project bases document creation on the preferences of large numbers of contributors by
codifying the basic actions of editing, discussion, and expressions of dissatisfaction into distinct actions that
can be handled properly in a large scale system.
3.1 Document Project Steps
The following is a list of the actions that occur during the creation of a document using the Document
Project:
Figure 7: Main actions that occur in the Document Project. Steps 1,2,3,4,5 and 11 can happen at any time
Continual Process
{ (1) User registration with username and password. If desired, the user enters in as much demo-
graphic info as she wants as so-called tags (age, gender, hobbies, occupation, etc.). Registration
with this limited information allows natural groups to form within the community of any given
document and the ability to recognize what what non-majority ideas are being disenfranchized.
{ (2) Document creation: any user can create a document that can then begin to be edited by
others. When the document is started, it is in Version 1. After each editing/voting cycle, the
document goes to Version 2, 3, etc.
{ (3) Document registration: users can join a document that they would like to contribute to.
{ (4) Listing and Viewing of documents. Figure 8 and 9 show examples of this.
{ (5) Viewing of statistics and history of changes that have been made in the document (see Figure
10).
{ (11) Allowing users to express dissatisfaction in the current version of the document (Discussed
in detail in Section 3.2),
During the Editing Phase
{ (7) Users begin the overall editing process by editing the document in a normal way, in that the
document is opened in an editor, and the user is free to make any changes and additions that she
would like (Figure 11). We will call this Edit Step 1 (of 2).
6See Rodriguez et al. Advances towards a General-Purpose Societal-Scale Human-Collective Problem-Solving Engine.
9
26. Figure 8: listing of documents that are being edited
After the regular editing in Edit Step 1, the user is then given a list of the changes that she made,
and is prompted to pack the changes in Edit Step 2. A package is one or more sentences that a
change was made in (including adding, deleting, and moving a sentence) that must all go into the
next version of the document, or all stay out, for the edit to make sense. Sentences in a package
do not need to be abutting one another, and they can pass between each other as well (Figure
12). Each package made will either go into the next version in its entirety, or not go in at all,
based on the support the package receives in the succeeding voting phase . A comment for each
sentence packages can also be added7
Between Editing and Voting Phase
{ (8) The list of packages created by multiple users are collected to be voted on (Figure 13).
During the Voting Phase
7The comments that can be attached to each Sentence Package are a way to have experts in a given
27. eld explain and teach
their rational for the edits that they make. Therefore, in order to bestow their in
uence, experts will need to educate people
not in their
28. elds by using comments that they attach to their edits. This can be a force that increases the knowledge of the
entire community, and is a way to continue to value expert opinions while at the same time stress empowerment and democratic
decision making.
10
29. Figure 9: the main document panel displays information and shows the users actions at any given time
{ (9) Sentence packages made in the editing phase are viewed by all of the contributors of the
document to vote and comment on if desired (Figure 14).
The user votes packages up or down depending on whether he or she would like to see that edit
in the document. Scalability to an unlimited number of users is attainable by not having each
user vote on every edit, rather by using a ranking algorithm that gives each user a combination of
randomly selected packages as well as packages that are becoming popular or contentious based
on voting patterns.
At the end of the Voting Phase
{ (10) The end of the voting phase, a method is performed to compile the document to a new
version. The method selects the packages that will be put into the next version of the document.
Two packages are in con
ict with one another when any sentence from one package overlaps or
abutts (i.e., there is no buer sentence between the two sentences) with a sentence from another
package. Packages that are in con
ict cannot both be put into the new version. The goal of the
compilation algorithm is to select the package set (a set of packages that can enter the document
with no con
icts) that gives the most agreement to the community that is writing the document.8
8As the number of edits grow to into the thousands, the amount of con
icts, especially for short (1 page or so) documents,
will scale proportionally with the number of proposed edits. However, the method is an attempt to have the least stringent
11
30. Figure 10: Example of viewing a previous version of the document. In this case, the document is in Version
3, but we are viewing the changes made from Version 1 to 2.
A score can be de
31. ned for any package or package set in numerous ways by using the voting
information for each package. One simple example of a package score is the amount of YES
votes it receives, and the package set score would be the summed total of the individual scores of
the packages in the set.
The search for the package set with the highest score can be performed in several ways. One
simple embodiment is as follows:
(a) Remove all sentences that have more NO votes than YES.
(b) The remaining packages are ordered into a list by their package score as described above.
(c) For each package in the list, remove the packages that are lower in the ordered list if they
con
ict with the current package.
The packages in the list remaining after this removal is the package set.
It is possible that this method will not
32. nd the optimal package set. For example, the highest
method possible of compiling a new version of the document while still maintaining document coherence.
12
33. Figure 11: Editing Step 1 (of 2) of the document.
voted package, if not put into the document, could allow for many other packages to enter in, which
would make for a higher overall package set score. A complication of the algorithm to search for
such scenerios is to replace the singe list selection (step b) with all N! (factorial) possible orderings
for the packages. Then select the list with the largest package set score. If the calculation becomes
too large (with N! needing to take place), a Monte Carlo algorithm can be implamented to search
for the highest scoring list and package set.
In either case, showing a graph of the package set scores of all possible package sets for a given
edit, ranked from highest to lowest package set score, one would see a curve that dies o as the
package set become less and less valuable. Only in the comparison of the N! package sets would
one be guaranteed to select the highest scored package set. The drop o of the ranked package
set scores can determine the level of overall acceptance of the version compile: a greater drop-o
means that the highest score package set is the clearest choice out of all of the other potential
package sets to maximize the satisfaction in the next version.9
After the editing cycle, the process returns once again to the Editing phase using a new version of the
document. The cycle of Editing and Voting then repeats.
9Since statistics will be available for the support level of packages during voting - if a package almost was chosen for a given
version, then the user can be prompted to enter a similiar package edit in the new version.
13
34. Figure 12: Creating packages. The fundamental unit of each package is the sentence (you cannot put half of
a sentence in a package, for example). Note that each package (1,2, and 3 in this example) can be multiple
sentences long, that the sentences in packages do not need to abutt, and, as seen in package 2 and 3 in this
example, can . Also note the user comments that were made for the packages created. These packages are
shown in packaged form in .
3.2 Dealbreaks
After any given voting phase, when a new version of the document is compiled, there is a strong possibility
that groups with ideas that are not in the majority10 have been disenfranchised, and handling this will be a
very important step in the creation of the document.
The diculty of reaching consensus between dierently sized groups can be shown in an example: there
are 100 people of persuasion A and 1000 people of persuasion B of a given topic. There is some change in
the document that is voted on. There are 10 votes not to change from group A, and 900 votes to change
from group B. This is a clear signal to make the change. But what if there is 70 votes from group B not to
10The de
35. nition of a majority viewpoint can be any idea that has the support of more than 50% and less than or equal to
100% of the people contributing to the document. Setting the number de
36. nition of majority to a high number, say 66 or 75%,
is seemingly helpful to create majorities that include a maximal amount of support, but they create impossibly high barriers to
voting in changes, and do not solve the fact that large groups of people will continue to be in a non-majority opinion, something
that must still be addressed. Therefore, any de
37. nition of majority is acceptable for this discussion, and we choose 51%
14
38. Figure 13: List of sentence packages that are ready to be voted on when the voting phase begins.
change and 400 votes to change from group A. Clearly, the majority vote says to make the change, but 70%
of the non-majority group didn't want it. This type of problem will arise naturally and often. One way to
solve it is to normalize the votes to groups. This is dicult and undesirable because the idea of a de
39. nable
group A or B or any other group, although useful, either doesn't exist in the real world or it is very dicult
to de
41. ling people. We are all individuals, act and think dierently, and are
always growing and changing. It would therefore be unrealistic to create these groups once or try to change
them before every voting phase.
The Document Project attempts to solve such problems by allowing contributors to mark particular
sentences in the document as dealbreaks: the contributors will not support the the
42. nal document if the
dealbreak sentence or sentences stay in the document (See Figure 16).
In any version of the document, it can be seen how many contributors are supporting the document
(could sign the document as is) or not supporting the document (have one or more dealbreaks that disallow
them from signing the document at the end) (Figure 17).This information can
uidly change from version
to version as new people enter the document, edits are made, votes are taken, new versions are created, and
new dealbreaks are expressed. (Figure 18).
User support and dealbreak sentences can also be seen on a tag based level (Figure 19), which allows the
contributors of the document to see when and how non-majority groups are being forced to not support the
document.
Using the information about dealbreaks and loss of certain non-majority groups from the support of the
document, one option for editors on any given editing cycle (in addition to adding brand new content or
making stylistic changes) is to try to come to consensus with a given dealbreak sentiment. During voting,
users can see that a given package they are voting on is changing a dealbreak sentence (Figure 20). Therefore
15
43. each voter has the power to guide how consensus is built in the document.
The premise of the deal break system described above is that if dissatisfactions from non-majority groups
are seen clearly by the larger group, and the loss of support is realized due to the dissatisfaction of non-
majority groups, then there will be an eort to come to consensus to keep the size and diversity of the
supporters of the document as high as possible. This is an optimistic idea about the innate desires to
understand and come to consensus with one another, but also, in many situations, decisions or causes made
without a large diverse group of people backing it will not be as in
uential in a pluralistic civilization.
Alternatively, a decision or cause that does have support from people with non-majority ideas is a very
powerful force for change, especially if it is an issue that requires support from a diverse group of people.
The goal is a maximization of people that support the document with the most diversity possible, and to
avoid gridlock when the system veers into a stando position.
4 Sentence Based Structure
The schema to store the information about the documents, their version histories, about peoples edits on
them, the creation of packages, expressions of dealbreaks, votes on speci
44. c packages, and commenting which
allows for the functionality mentioned above is novel. Although some of the functions mentioned above
could be implemented with straightforward use of traditional di and merging algorithms (such as the
compilation phase), the totality of the features mentioned above, could not be performed without a new
methodology. Moreover, an additional feature of multiple language integration (discussed in Section 5 could
not be performed either.
The fundamental unit of the new method is the sentence. In the beginning of each new version of the
document, each sentence is assigned a location: set of 3 numbers similar to the Dewey decimal system that
explicitly de
45. ned what the Chapter, Paragraph, and Sentence location is for that sentence11. For example,
the
46. rst three sentences in Chapter 2, paragraph 3 would be C:2/P:3/S:1, C:2/P:3/S:2, C:2/P:3/S:3. When a
document is edited, sentences can either be deleted, moved, changed or inserted. These actions are logged for
the aected sentences as CHANGED FROM, CHANGED TO, DELETED, INSERTED, MOVED FROM,
MOVED TO. For the sentences that are CHANGED TO, MOVED TO, and INSERTED, a location is given
that is fractional to the origional location structure, de
47. ning where the sentence exists with respect to the ori-
gional document sentences (Figure 21). During the edit, new sentences are created, along with the fractional
location information of where they exist with respect to the origional version. If a sentence is CHANGED TO
or MOVED TO or both, this adapted sentence has a pointer to the sentences that were CHANGED FROM
and MOVED FROM respectively. And reversely, the CHANGED FROM and MOVED FROM sentenced
point to their CHANGED TO or MOVED TO counterparts. This system therefore allows for complete
tracking of the creation, evolution, movement, and deletion, of every sentence as it is manipulated from
version to version.
The de
48. nition of con
icting packages (discussed in 3.1) falls out from the sentence locations of the
proposed changes: if any two packages have a sentence with a location code where the Chapter and Paragraph
numbers are equal, and the Sentence number is equal or less than an integer value away from the other, then
11We will assume here that the subdivisions of text that exist in the document are chapters, paragraphs, and sentences,
however, any set of subdivisions can be used
16
49. the two packages are in con
ict. This keeps the criteria of having at least one sentence from the old version
between sentences from packages that are accepted into a new version.
When new versions of the document are compiled, the sentences from packages in the selected package
set that are INSERTED, CHANGED TO, and MOVED TO are all turned on in the new version, and
the sentences that are DELETED, CHANGED FROM, and MOVED FROM are all turned o in the next
version. The location numbers for each sentence are then refreshed by reassigning integer locations to the
sentences in the new version.
5 Language
As mentioned in Section 2, some of the most needed collaborations in this world involve people from dierent
cultures. By enabling people that speak dierent languages to back a common cause, traditional governmen-
tal and cultural boundaries become less of a barrier between meaningful collaboration between large groups
of diverse people.
Although the Document Project can be used in any single language, another aspect of the method is
its ability to allow people that speak and write in dierent languages to edit the same document with one
another. Because the sentence is the fundamental unit of editing, discussion, and voting, each sentence can
be immmediatly and automatically translated to other languages when it is created. Comments on packages
and votes as well as the user tags are also immmediatly and automatically translated12. Each sentence
is in reality a multitude of sentences, one being the origional, and the other being automatic translations
of the origional into all of the languages that the Document Project supports. By doing this, the actions
allowing collaboration, editing, voting, and discussion within the same document can be performed using
any language. This makes the entire process of collaboration on a document language-independent.
An example of this is shown in Figure 22 and 23. This particular document is being edited by English and
Spanish speaking people. As edits and voting take place, the website shows the steps in only the language
that is being displayed at the time. Astericks next to sentences and comments are used for all text that has
been automatically translated (Figure 24). After compilation, when new sentences enter the document that
were created in dierent languages, one can see that the new version has sentences that were voted in that
were origionally created in dierent languages (Figure 25).
6 Conclusion
There is little more impressive than diverse groups of people overcoming their dierences to express solidarity
for needed change in their world. One could argue however that the current tools that people have to
12This automatic translation is currently being provided by Google Translate: http://translate.google.com/. Machine transla-
tion algorithms automatically translate text from one language to another. Such algorithms are becoming increasingly accurate
(See http://en.wikipedia.org/wiki/Machine translation). One of the driving forces of automatic translation is to reduce the
communication barriers between dierent cultures. Current translation technologies focus on translating the sentence, in that
they do not chop the sentence up before translating, nor do the algorithms look at neighboring sentence's content when deter-
mining how to translate any given sentence. This is telling of the power of sentences to convey a single concept in all human
languages. This concept allows machine translation algorithms to be a good method of translation within the Document Project
system.
17
50. communicate with each other leave something to be desired in enabling the
uid creation and adaptation of
groups in ways that our world needs.
The Document Project attempts to allow an unlimited number of people with disparate ideas, back-
grounds, cultures, and languages, to coherently and sensibly interact - without leaders or representatives
- to create a petition, a course of action, or any other type of docuement together when the need arises,
and to maximize the in
uence of the interaction by making the contributions from each individual as self
empowering as possible.
The Document Project attempts to creep over whatever societal potential energy barriers exist for
needed change as best as possible. It does this by tapping into the greatest potential our community has:
its people, coming together against their dierences to support a cause they all believe in, and knowing that
the consensus they build is their fundamental power.
18
51. (a) voting on a package
(b) viewing a single package in the document
Figure 14: Example of a vote on a package created in Figure 12 and shown in Figure 13. The user can always
click on See single sentence package change in version to view the package they are voting on (b). In this
case, the user is voting to put this package into the next version, and is writing a comment as to why.
19
52. Figure 15: Example of ranked package set scores for a given version compilation. The algorithm that selects
the package set that will be used tries to
54. Figure 16: Example of a users dealbreak page. A user is allowed to express a dealbreak at any time, update
old dealbreaks she made in the past that may have been resolved, and express new dealbreaks from the
current version of the document.
21
55. (a) contributor details
(b) dealbreaks for entire document
Figure 17: (a) View of contributors of the document, whether they are supporting the document (could sign
the document as is) or not supporting the document (have one or more dealbreaks that disallow them from
signing the document at the end), and the words (tags) that describe them. The view shows what group
the user is currently in based on the dealbreaks they expressed (Figure 16). (b) the dealbreaks for the entire
document are shown as the reasons why contributors are not supporting the document.
22
56. (a) Version 1 (b) Version 2
(c) Version 3
Figure 18: Example of the changing of contributors as new people enter the document, edits are made, votes
are taken, new versions are created, and new dealbreaks are expressed.
Figure 19: user support on a tag-based level.
23
57. Figure 20: During voting, users will know that a given package they are voting on is changing a dealbreak
sentence.
24
58. (a) original version
(b) new text added
(c) di showing sentence locations
Figure 21: (a) In the beginning of each version, there is fresh set of sentence locations. When a change is
made (b), the adapted di algorithm assigns of fractional locations to new and moved sentences (c).
25
59. (a) Spanish view
(b) Engish view
Figure 22: The homepage for a document shown as both English and Spanish. Note that the user tags of a
potentially bilingual group are completely converted into the two languages respectively.
26
60. (a) Spanish view
(b) Engish view
Figure 23: The document displayed as both English and Spanish. Note this is Version 1 of the document,
and there is a (*) in the English version on every sentence, but there are no (*) in the Spanish version. This
means that it was originally submitted in Spanish.
27
61. (a) Spanish view
(b) Engish view
Figure 24: 4 four sentence packages made on the document in Figure 23, two from a person writing in Spanish,
and two from a person writing in English. Note the changing asterisks noting automatic translations in the
two language's views
28
62. (a) Spanish view
(b) Engish view
Figure 25: Version 2: In this example, all four sentence packages shown in 24 were voted into Version 2.
Note the reciprocal nature of the asterisks which denote automatic translations now integrated into the new
version of the document.
29