Forrester's Law, the conflict-driven view of wiki and Lowy's Law all suggest that dialectic is more important to evolve wiki as a concept, than agreement. A number of issues in wiki management and wiki governance are poorly defined enough that known trolls continue to debate them. The following summarizes such a dialogue between two such.
Craig Hubley claims that "wiki users are by definition willing to agree or refine 'consensus' as they go along since any user can challenge any word on any page. All of them have to evolve what they mean by consensus as they get new participants, they have to develop politics. Wikipedia had to develop ArbCom."
David Mason suggests that this is not consensus in the political sense due to participation inequality: "projects like wikipedia basically have a very small percentage of people editing documents, everyone else goes away when they realize what its really all about." Unlike voting or even use of blogs or email, "my experiments on wikipedia, for example, were mostly just plowed over... its basically another group of people writing history according to their rules... everything is a "chilling effect" when its attacking one of their pet interests." So the high barrier to participation and dominance of technocratic cliques, in Mason's view, prevents any real consensus in the political sense from developing.
Hubley doesn't necessarily disagree but notes that even voting and speaking out during high-stakes elections for public officials have very serious participation inequality problems, such as the old voting more than the young, and English native speakers having far more voice than immigrants, and very few people being able to raise the funds to run. So the problems of wiki are not different from those of politics as usual, and those are not impossible to understand or manage. Though Hubley would be first to list a lot of problems with wiki applied in politics:
he can also see a way to solving some of the problems:
Wiki governance should be about more than infrastructure owners trust: who the domain owner personally likes. Wikipedia was preceeded by bbc's hhgtg and everything2, and many other attempts to assemble encyclopedic databases from many users and apply mass peer review. Mason takes this as evidence that "wikipedia 2007's data format" could be replaced, which Hubley doubts: "HHGTG and everything2 are toys, they just aren't widely mirrored nor consulted by the average person, unlike Wikipedia; they don't work in many languages and they didn't spread their toolset to many other users. They just aren't comparable to the spread and success of Wikipedia." Though it might have been otherwise, it wasn't, and data has weight.
Mason admits he's glad "everything2 didn't "win," since its point of view is heavily geeky, but hhgttg "winning" might have been better than wikipedia, perhaps skipping ahead when considering some issues around editing and points of view. Maybe the BBC would be a better guardian than Wales." Hubley agrees and notes that Wales serves only one purpose at present, that is, preventing mediawiki coders from automating functions that should remain social and performed at the initiative only of real people (like greeting). He also notes that the BBC and Wales have been talking from time to time and that Wikimedia has for a long time considered acquiring an advisory board. Once the wiki technology is familiar to librarians, journalists and academics, recruiting a good board might be easier.
Mason admits its exceptional status and "considering that wikipedia only involves a very small percentage of people on an ongoing basis, its remarkable. But wikipedia is going to change too, and probably its content being under the GFDL is more significant than its current mechanisms and form." Hubley agrees and points to the ubiquitous phrases "GFDL corpus" and "GFDL corpus namespace" to neutrally describe all the texts under that license, and the names that seem to all be ultimately influenced or decided by Wikipedians.
Mason and Hubley agree that Wikinfo is very strategic and that "sympathetic point of view" at least as an option solves many problems. Hubley believes that it wouldn't work without Wikipedia: "there still needs to be a default messy place where all the POVs meld into a troll war, which is what Wikipedia is: "battlefield of ideas". Wikinfo wouldn't work without Wikipedia and the fuel of hatred that makes trolls attack each other to take over pages... but when it's obvious there's no way to resolve the dispute about naming things, then it helps to have Wikinfo there to let the case FOR a thing be stated clearly." Mason is less confident that the conflict is necessary and suggests that the views may never clash "or even not merge into a troll war, specific groups can import the wikipedia corpus en masse, and edit as required to represent their own point of view. You don't have to reconcile everything, but you can use it to know where you stand with a particular group, by definition and in particular difference. If the Mennonites want to be 80% wikipedia compatible, and 20% Mennonite compatible, great." Hubley called this an epistemic filter in his 1993 contributions to the UN University Millenium Project, which was very extensively quoted in its 1993 State of the Future report. For fifteen years Hubley has advocated that each group be able to apply its own causal matrix to a common map of testable facts and provable correlations.
Mason suggests the most useful tool would let his mother " tell her stories of trading with natives in Saskatchewan in the 40s, in a web of experiences, viewpoints and data. Perhaps it's wiki gardeners that need to be encouraged the most." Hubley was working on a similar project for psychiatric survivors very recently and notes that one US military project in the 80s did exactly this for soldiers wounded in Vietnam and their families - documenting each incident of harm as an entire hypermedia matrix of experience. Public wikis however require people to deliberately protect the more introverted or less popular from those who would oppress them with the dominant POV: "ethical trolls protect gardeners in a way that sysops do not... making room for minority views including unpopular ones. If radicals have no home then individuals have no rights... as usual in politics..."
While Mason and Hubley agree they are both "big advocates of trolls" and prefer "troll-friendly" wiki management and governance approaches, Mason thinks Hubley needs "to reconcile the notion of trolls with your top-down, kill anyone who's not part of a central point of view position." Hubley admits that he does in fact believe that naming conventions and other syntactic standards are absolutely required and that chaos will reign without them, and that exclusion to the point of violence (e.g. lawsuits) might well be required to prevent serious loss of data integrity. He however takes the classical liberal position that without providing some common infrastructure and some law-like protections, the trolls not only cannot get started but would be reliably hunted down and either excluded successfully or killed. As one example, the Chinese government would love to find everyone who'd ever criticized it online, and know who all their friends are. This is a strong argument against using transparent social networks and real body identities.
 data integrity
Mason finds the lack of automation that the social view of wiki implies, tedious: "I find some operations, such as renaming a page... if someone misnamed something as it is defined... to be not at all as easy". Hubley thinks this should be easy if and when the movement is towards a convention not away from it, but doesn't not want the tools to help anyone "reliably propagate a bad naming CONVENTION. Everyone is sometimes wrong. When we are wrong about the same thing twice, we become stupid. When we defend our stupidity by making up "reasons why" post-facto that bear no resemblance to our pre-facto rationale, then we are no longer saveable and have become compostable."
However, this applies only to highly automated bots or many-pages-at-once change. Hubley and Mason agree that ordinary users not using any tools should be able to submit, in Mason's words, "valid but not perfectly formed content, and then reforming it to fit the wiki's notion of order. That's perhaps the only way to make it truly inclusive, otherwise it's only suited to the indoctrinated." Hubley notes that this is actually exactly how well run wikis work: even if content is poorly formed, it is quickly "wikified" by more experienced users, who indicate in edit summaries and talk pages why they made the changes that they did, and who scrupulously avoid omitting any claims, references, sources, links to other relevant pages, and will confine their changes to copyedits and link fixes and so on, only changing the POV very slightly. People are however encouraged to restate their own contributions to prove that they understand how to move towards neutrality or at least balanced multiple views. This accomodates, in Mason's view, "someone who wants to share their point of view but can't do so according to particular conventions and sensibilities." Where there is no special power-over each other, Hubley and Mason agree that technocratic cliques should not be excluding others. Though Hubley notes that mastery of a toolset that makes page name changes very easy would provide great leverage to the technocrats, and suggests this is not a priority when considering upgrades to wiki software, for this reason. He also has a bias towards users who take the time to get the naming conventions right initially: "I agree some common transactions need to be simpler, notably correcting inbound links to clearly wrong names, but it's not terrible to assume that page renaming is a high overhead serious pain in the ass, and therefore you have to get it right from the beginning. Making it too easy would encourage sloppy naming, just as it does in people's home filesystems."
Mason finds this "a pretty crappy solution compared to the advanced facilities available today. I suspect the difficulty in renaming has more to do with the inherent design of wikimedia." That is, with relying on social rather than automated reference to ontology that already exists. Even edit conflicts are dumped into each individual user's lap to resolve, without an attempt to merge the pages together. Hubley finds this acceptable since each author should have the chance to review the other's changes, and an opportunity to get together in other media like chat or email to debate the changes and conflict. It's a social opportunity, not just a database integrity problem.
While Mason finds all this "probably a sign of feebleness" and suggests Wikipedia may be "only interesting due to its scale", Hubley argues that this is the only thing that would ever make any wiki at all interesting: "The scale is what makes it robust. Also what makes it persistent. No one is going to redo Wikipedia to do it SLIGHTLY better, they'd only redo it to do it LOTS better. And that's unlikely within our lifetimes..." As in the 1980s when there were many hypertext systems of very high quality like KMS, IBIS and NoteCards, but none achieved any critical mass of users, and were replaced by HTML, the scale will make a simpler system replace complex ones.
So while people will try things and some of them will be incompatible with how wikipedia works, they'll be faced with the sheer weight of Wikipedia's database, persistence of its naming conventions progagated over hundreds of mirrors in dozens of languages. And of course the inarguable fact that "Wikipedia, by its own definitions, is - editable by anyone, without registration". According to Mason, "this makes it broad and unreliable, an excellent starting point for important information, or good enough for trivial information. Nothing else." Hubley takes exactly the opposite view: "this makes it diluted but potentially more trustworthy than any other source, at least on pages which have received review from major scholars or their students, who are welcome to log in as themselves and verify certain versions of pages. No other source can claim that everyone in the world had a chance to edit, but no one chose to. That is quite an argument in favour of the information being right."
Mason and Hubley agree that in practice this degree of mass peer review doesn't happen. Mason says "it is ruled by a very small percentage of users, and many common facilities are clunky, at best. So it doesn't really meet any goals, though it does have a lot of good content." Hubley says "it's a working prototype of something that would probably have simple tools to correct things like spelling errors or malformed links, and more complex ones like interpreters that watched for causal language ("since", "so", "because", "thus") and perhaps applied some epistemic filters at least to add tags to attract other users to review it."
 diffs, trolls, factions
Hubley claims "it's the diff between Wikipedia and Wikinfo that's interesting, just as it's the diff between versions that tells you the most about what is
going on within either... where the troll fights are..." Mason notes correctly that this is only of interest if consensus on the issue and uniformity on the page content are somehow to be seen as equivalent.
He notes also the participation inequality issue, by adding (three notes) to a quote from Cory Doctorow: "Brittanica tells you what (a few) dead white men agreed on. Wikipedia tells you what (a few) live Internet users are fighting over." (imagine a sissy slap fight, with the winner the one that didn't walk away)". Also that Doctorow himself will "post about you and your 'chilling effect' on boingboing's front page" and accordingly is clearly willing to use his own name and prominence to defend the winning faction of trolls which (him being a technocrat) will include himself more often than it will include his opponents.
Hubley finds this just inevitable: Doctorow is a so-called smug pro-trolling troll: "That view is where "troll-friendly" comes from. In a fight each side calls the other "trolls" so unless we are all trolls, we've just handed one side the keys to lock out the other." Doctorow at least wants the fight to continue, rather than to gain the keys for himself.
Mason asks wrly "do trolls fight to the death?" as a way of challenging the metaphor Hubley and Doctorow both use, and the conflict-driven view of wiki.
 POV nesting and social networking
To minimize conflict and align goals of each wiki better, Hubley advocates, in addition to the Wikinfo "sympathetic" view layered on the Wikipedia "neutral" or "truce" view, an "additional layering/nesting/cascading to import to a consortium (which would be "even more sympathetic"), then to private projects" as his primary priority. He's quite a getwiki fan, too, but can't use it because it's single-sourced software.
Hubley says "the fact that each wiki is aligned to one set of POV assumptions that makes them all useful. Without that, none of them would be useful. And that's
why interwiki is not useful." Mason sees them as a more semantically exact form of what's already done: "I see people refer to wikipedia articles using ordinary hyperlinks all the time, in email, irc, the press, you name it. In the end, interwiki links are just a short form of this convention, which is all a basic layering of the concept of hypertext" similar to say Ted Nelson's Xanadu, the first robust description of a hypertext layering system. Hubley notes however that it's what's unique about Wikipedia that makes it worth referencing, not the fact that it's a wiki, so there's no reason why other wikis should be just as easy to reference while, say, academic papers aren't. The wiki-ness is "not where the store of value is, that's just the way the medium of exchange spreads. Interwiki is wrong to assume that "wiki" has the value of "pedia", it doesn't, the wiki is just a front end. The pedia is the value." Mason doubts this and claims "It's understood these days that what's important in social networks are the connection points. People surf information via connection points, they work across languages. So the connection points (links) are as important as the information. If no one pointed you to the information, it would just sit there." While Hubley agrees with that, he doesn't think the social networks are anywhere near as valuable as the content for the reason Mason himself suggests: too few people are participating. So while you might learn a great deal about the tiny minority that edits wikis under their own name or a consistent pseudonym, that would not be anywhere near as useful as content all meeting a very narrow set of POV and syntactic conventions. Also "while the user pages of Wikipedia are one of the most reliable and robust social networks ever created, you don't see anywhere near that value in other wikis".
Hubley says that the conventions determine the value: "if interwiki worked only for mediawiki it would be much more meaningful because at least mediawiki users follow conventions set by Wikipedia. Trying to make links to all kinds of articles everywhere look the same is just adding noise, via over-egalitarianism." Mason counters "there is a value to relevance expressed to you via human networks and particular points of view" but Hubley sees no contradiction: a human can express their POV on a talk page (attached to the topic) or user page (attached to the account) without having to "own" the topic or rewrite everything said about it into a custom article or editorial. In other words, "why not just edit the Wikipedia article and point others at your version?".
Mason suggests that the semantics even of Wikipedia are weak: "HTML defines a very small set of conventions, everything else is practically semantic mush at this point, unless we can get a handle on rdf type syntax, any new definitions in wikimedia are going down this path. This is happening now in wikimedia, but it is not digestable by common mortals. Maybe visual controls will be the answer." Hubley sees no value in visual representation of semantic mush but would like "numbers on things like which links are most transited, just as you get for web ads. It's ridiculous that the porn 'industry' and web spam have far more information about who clicks on what than the authors of articles about everything read by everyone. Wikimedia is really failing us all here."
But even so, it's doing better than most other wiki management, as "most of those wikis don't have any user critical mass nor citation discipline, for instance, so you don't have any reason to believe it's more than one person's opinion." Mason agrees but he sees Wikipedia as having only slightly more mass on most topics, rather than "an encylopedia in the classic sense" which would have more discipline to review each topic and provide editorial balance among the topics.
He sees the benefits of something like Wikipedia "in terms of global consensus and information relation" rather than just in the fact that many people use it.
Hubley would rather just put numbers on all of this and stop arguing about it: "many of those objections go away if the number of persons who regularly edit who have looked at a page were known... if they looked and didn't edit, then, that can count as some evidence it's more reliable... Also add in the citation index of any references that stand the test of time... to weight academia properly. Either would quantify the advantages Wikipedia has... letting us start to identify who else is providing a useful POV discipline." Interwiki without that is "just useless, a spam generator like the blogosphere."
Mason suggests that dropping interwiki "still leaves specific knowledge that has no place in wikipedia, and these ordinary conversations are really much more important than wikipedia. They are people's actions in life, not what they are referring to in history, theory, or "fact" (how many people have been told they would die in 5 years, or fail in life). An email, or your despised blog entry is an exchange on these lines, not an appropriate wikipedia entry." In other words, all that information that is "not encyclopedic" needs somewhere to go, including what's on your own hard drive, iPhone/iPod, or scribbled in your diary.
Hubley notes that this can be solved by nesting all the way back to the individual POV on your "very personal computer", and that "if you want to share and commit to agreeing with others, just state that you'll maintain no page on this topic on your own but only a common version with specific others. You can always change your mind and fork - forking is easy. But merging is very hard, and we should therefore err in favour of everyone working from one version all over the world. So Wikipedia is the logical starting point."
 format durability
Mason notes the history of data storage and says "there's no saying one format or another will be a definitive format into the future" but Hubley notes that the syntatic, rather than the physicl medium, has been quite stable, and asks if "at the height of the Roman Empire, after the Greeks and Romans brought written language to so many people and ruled them by it for hundreds of years, the fact that in two thousand years we'd still be using the Greco-Roman character set was not predictable? I say it was." Mason answers with a chunk of Greek which Hubley can't read, but only because it's in Greek. The letters are clearly recognizable as pi, omega, alpha, sigma, etc., which have exact equivalents in Latin and thus English. Mason even typed them on his keyboard, which suggests that the symbols have survived many physical media shifts from carving in stone with steel chisels to Unicode on AMD processors. Hubley claims that "the character set was demonstrably much more efficient than competitors like for instance Chinese which takes a long time to learn. So once you have a widespread standard that's efficient enough, you can count on it sticking around for a loooooong time." Mason responds by quoting some Latin, which again Hubley can at least sound out... though he can't catch the intended meaning, he could repeat it to someone who spoke Latin.
Hubley does admit that "we aren't using their" Roman "numbers", but those were, he says "demonstrably less efficient than many alternatives..." while the alphabet was not. So efficient that mathematics adopted the Greek alphabet to mean the same thing all over the world, and an equation would be relatively understandable (at least syntactically) to anyone who was reasonably involved in a discussion on a topic where an equation would be considered normal to use.
Regarding hardware, Hubley says "we still build railroads on the Roman gauge that was used for cart tracks" while Mason "spoke to an engineer cursing that fact, the narrow width is the cause of many an accident and frustrating speed limit. I wonder who in the future is going to be expelled from society because their point of view doesn't jibe with wikipedia." To which Hubley responds "only trolls". Thus, it's the expansion of the definition of trolls such that all users are trolls that's of concern.
Mason finds syntactic translations to be misleading and often meaningless, e.g. "the MS Word 95 document semantics could only be translated into a standard format using format-it-like-word-95-bold-italic tags (or balloontext it like a Word 2007 inline comment?)". He argues that information has been lost in any such translation. Hubley however takes this as evidence that Wikipedia would be resistant to change, and again notes the data weight: "The amount of globally interesting information in MS Word form is very high, but the amount of globally SHARED information of that kind is very low. I'd argue Wikipedia is already bigger than the latter. HTML still dwarfs either" but has no POV or cite discipline, and not even any truly semantic links.
Mason considers this irrelevant as "the point is sometimes information is lost during a transition. Shrug. Information is lost every time a mere mortal tries to contribute to a wikipedia entry." Hubley disagrees as even failed tries stay in page history. He notes that well-run wikis bend over backwards to ensure that the number of actual lost transactions is minimized, restricted usually to spam and to copyvios, that a balance/hide/ignore/delete protocol tends to radically minimize the amount of actual deletion.
Mason finds moderation of any kind suspect; He notes an event where he "was participating in a Globe and Mail conversation (I get posting sickness), only to realize they were retroactively removing comments as the conversation progressed, which isn't really "semi moderating" as they called it, but rather "active point of view manipulation."". Hubley finds this a good reason to avoid web threads and mailing lists, which "due to artificial scarcity, tend to require their administrators to aggressively forbid certain topics or exclude users who want to drone on and on about one thing or with an unpopular view. A wiki, by contrast, can let all that stay in the page history of a few pages and let the view finds its own friends." Because moderation bias is inevitable, Hubley considers wikis inherently more honest: "wikis, which advertise up front that they'll "Be bold" about editing, are much more likely to attract ethical participants who want the rules to be announced before the fact and kept to." Hubley and Mason agree that "the really meaningful tacit rules" are unadvertised, but Hubley thinks that's just part of being "social".
Hubley and Mason agree that the main problem is "wikipedia thugs waiting to erase your point of view from their favourite pages" so that "no one will ever see your edit." Mason's solution is to have more inter-user respect or status or sanctity of a user's exact words, but Hubley's solution is more trollish:
"1. You can recruit your own friends to fight back as trolls, and they can keep coming back even if "banned for life", because Wikipedia is so deliberately leaky"
"2. No one ever removes the page history so anyone can find and restore your edit if they find a page too POV"
"3. There are various ways to harass the thugs using bureaucracy, e.,g. ArbCom, label them as sysop-vandals and so on, which have proven effective in making Wikipedia "not fun" for them to continue to control"
"4. No one group, not even Jim Wales' best friends, is really capable of continuing to dominate the discourse just using infrastructure owners trust (i.e. the Globe and Mail), and no one is being paid to moderate stuff so at least there's no money skewing the moderation..."
Mason finds this a very high overhead conflict-prone solution: "I have a life...I think the only solution is augmenting wikipedia." Hubley however "can't see any way of augmenting that would not enable current dominant cliques to extend their own power, just as any advance in weapons that require training is an advance in power for the rulers, not the dissidents. Eventually the dissidents become suicide bombers if public investment is going into bombing from above."
Hubley and Mason despise PHP and consider mediawiki to be problematic for using it. Hubley thinks Java wiki authors "should be scheming to replace mediawiki, and convince wikipedia to dump PHP in favour of a JSPwiki supporting the same format and social process... which would be unreasonable if it wasn't JSPwiki's only and native format, so wikipedia users could be sure the effort would remain focused." Mason agrees but doesn't see it happening, as "the author of jspwiki, Janne Jalkanen...has made it painfully clear he is utterly uninterested in being mediawiki compatible. And I think that is fine, though he's a bit harsh in stating it." To Hubley this harshness, which consists of calling the author of the idea a "troll", proves only that the troll is right, Jalkanen wrong: the idea and its responses.
Mason notes that "There are at least five wikis I know of written in java. The only true "java wiki" is JavaDoc's tags." He finds wiki creation easy, and notes that his "own banana wiki created in an evening a few years ago had some great features, and a very 'a-pealing' name." Hubley's experiments with custom wiki were also fruitful but stalled due to lack of data mass and peer review. Accordingly he is inclined to "not believe that anyone else is going to get further than I did trying to justify custom wiki code with a custom data format. Like it or not, mediawiki is the standard. I see a lot of serious players saying that."
He claims "Wikipedia will still be around in 10000 years, sadly perhaps. But over that time it's technology base and user interface will change probably a thousand times." He therefore favours the terminology "GFDL corpus", "GFDL corpus namespace" and perhaps "Wikipedia tags" and "Wikipedia user network" to more clearly separate out what the deliverables and components are: "There's also a growing CC-by-sa and CC-by-nc-sa corpus, to which some Wikipedia tags and user networking could apply."
Mason notes that a construct only holds meaning if "precisely defined," and if they are "it could be imported reliably. All the non precisely defined parts will fall off." Hubley just doesn't believe that and argues that "poetry is a counterexample. Much of the important information in a culture is carried by very subtle nuances, associations, and shadings. That may be true of punctuation as well... I am saying, bluntly, mediawiki format and English itself will converge. Each will lose a few things." Mason says "There'll be english used in rap songs, signing, english used on cell phones, chat boards, talking to family, employers, the church, strangers, drug dealers, doctors, the list goes on" which may diverge from Wikipedia's usage: "I can easily see JimmyWalesKeeptheSysopsHappy forms of communication, which no one uses at the dinner table." Hubley agrees but "that would just mean that mediawiki format expands to support all those means of expression, not all of which get into Wikipedia, but all potentially could. There will be at least one good example each of rap, signs, cell phone dialogue, chat dialogue, etc., in Wikipedia or some other mediawiki, and if that requires certain markup then mediawiki will support it." Mason thinks however it can't catch up: "The reality is reality doesn't converge, it disperses. Maybe computer mediation will be helpful here, with a success rate on translating swahili, sms messages, and spoken text in a conversation roughly equal to modern OCR or speech recognition (eg, not good enough, unless you are looking for comedic value)." Hubley disagrees flatly and points out that "given Wikipedia in English and Wikipedia in Swahili, all cross-linked as they are now, I have a terrific basis to translate nouns, even very abstract ones a typical translator has never used. That isn't everything, but it sure is s start. Again the links between versions are adding value that isn't there in less disciplined data like interwiki or web threads."
Hubley also finds efforts to standardize wiki format to be wrongheaded, as they try to lurch Wikipedia away from its core process adding value: "Creole won't even be in there unless it gets exactly in between" English and mediawiki "and anticipates shifts in" these "that make the two of them more like each other" and also more like existing formats "like W3, XML and HTML."
Hubley thinks Java wikis are "presently not in the game as it fails to respect the conventions of English or Wikipedia and seems to want to try to tell us English data should be named like Java data, which I find is utterly wrong... for the reasons I never liked Java in the first place... simple stupid hierarchies. It forces a taxonomy when it needs broader ontology, and it needs to respect that code is just data from a different audience, as both LISP and even twiki do", in other words, they support reflexive definitions of code as data, and ease the transition from using wikis to programming wikis, since the conventions don't change.
Mason sees no overt link between the data format and the code: "jspwiki at its basis has nothing to do with java any more than mediawiki has to do with php.. maybe they will allow inline java/groovy scripting for logic, that'd be great. It is its own thing compared to mediawiki." Hubley finds this not even reflective, and sees no value in "requiring people to learn three incompatible languages (PHP, scripting, mediawiki data format) in order to actually control the entire mess."