Document Mode Considered Harmful

The ultimate WikiHeresy?, but one that I think is becoming more true as WardsWiki continues to (d?)evolve. For a long time, it has been accepted that DocumentMode is the preferred format of pages (and that ThreadMode pages should be the means to that end, and nothing more). It has long been held that by this process of continual ReFactoring of conjecture by a group of intelligent individuals, an authoritative consensus - those pages in DocumentMode - would arise on this site; and that WardsWiki would be a shining beacon for ComputerScientists and programmers 'round the world.

ThreadMode, of course, has long been held to be a lesser form of Wiki content, one whose primary purpose is the creation of DocumentMode content. Many consider (or at least did so in the past; I'm not sure how true this is today) ThreadMode content to be inherently transient in nature - it has long been held that ThreadMode should eventually be refactored into DocumentMode, or else deleted.

Bollocks, I say. Horse manure, scat, nonsense, rubbish, and BS. :)

Why do I utter this heresy? For several reasons:

For one, the claim that this site is (these days) authoritative on anything is patently absurd. Perhaps in the days when WardsWiki was the hangout for the XP and DesignPatterns communities, this site was authoritative on these topics - being a gathering place for the leading researchers and advocates on these subjects. Currently, however, WardsWiki is at the leading edge of nothing. And even in the "good old days", many extrapolated leadership in DesignPatterns to leadership in ComputerScience in general - and thought that this site was (or should be) a website of record on numerous computer science topics; and that the DocumentMode content was the repository of this record.
For another, DocumentMode requires a community consensus - a GentlemensAgreement - which allows it to be produced. This consensus must cover numerous areas, such as how and when to ReFactor and delete content. When the site was a much closer-knit community (with WardCunningham an active participant), that consensus was reached. These days, the original members are mostly gone; there is no single unifying topic for this site, there is a noticeable lack of trust among many of the members of this site, and Ward seldom participates here anymore (and I can't say I blame him). The GentlemensAgreement no longer exists. Instead, we get EditWars as various factions try to claim DocumentMode status for their pet rants and arguments - to claim this site's "stamp of approval" (to the extent it exists) for their opinions, and to deny the same to any competing positions. Attempts to ReFactor out chaff in order to leave wheat further undermine the consensus, as disagreement arises over what constitutes wheat from chaff. In some cases, the science is well-established enough for an independent (and trained) observer to tell the difference; in other cases it is not.
For a third, the procedure by which DocumentMode is arrived at - community consensus and refactoring - doesn't always produce the "truth"; in many cases, it represents little more than the collective opinion of the participants here. In many cases, that is congruent with established doctrine - the participants being trained scientists, and relaying the information that has been established by the ScientificMethod. In other cases, DocumentMode is little more than GroupThink that has been given a stamp of approval by virtue of everyone present nodding their head. For a long time, this site advocated opinions on matters such as DynamicTyping, SmalltalkLanguage, etc. that were definitely not shared by the ComputerScience community at large. In many cases, that was because advocates of competing philosophies didn't participate here (or did so in much smaller numbers). (This is one reason that I consider TopMind to be a useful participant here, despite some of his behavior - precisely because he rattles the cage).
Finally, DocumentMode by its nature is static; many consider the ultimate form of DocumentMode to be a page which becomes effectively ReadOnly? - one that is sufficiently complete, accurate, and authoritative that it no longer need be edited. (Many such individuals argue against RecentChanges and the like, because DocumentMode pages generally disappear from any list of actively-edited pages). The sort of page for which DocumentMode is appropriate - established doctrine and advice which reflects the consensus of the community at large - makes this a compelling argument, as such content is perceived to be of high value. However:. Such information almost always exists in more authoritative forms elsewhere. Papers and journal articles. Textbooks. Online lecture notes. Even WikiPedia. Other than the instances (such as XP and DesignPatterns) where this site represents the LeadingEdge?; any DocumentMode content to be found here will likely be a) redundant with, and b) less authoritative then, content to be found elsewhere. In many cases, the most useful DocumentMode pages are those containing forwarding links to some other site, written by someone who knows (better than the folks here) what they are talking about.

So, if this site is less authoritative than others on the Web (which it is), and attempts to direct this site into a sort-of WikiPediaLite? (ignoring this site's founding status in WikiDom) are unlikely to be successful - then what is the value of WardsWiki?

Answer: The very thing that has long been held in ill-regard - ThreadMode.

The most compelling, interesting content on this site isn't the stuff reachable only through FindPage; it's the stuff that shows up on RecentChanges. It's the conversations that people have with each other (on ComputerScience topics; I'm ignoring the issue of OffTopic content), the expression of opinions (valid or otherwise) and the challenges to those opinions (again valid or otherwise), and the interplay between the participants. It's the dynamic content that makes this site valuable; and that cannot be replicated (easily) on more document-centric media like journals and textbooks.

Some may argue that ThreadMode isn't authoritative; that a visitor here will have a hard time discerning TheTruth? from an extended conversation/argument between numerous individuals on a given topic. Read the above: This site isn't - and shouldn't attempt to be - an authoritative reference on the discipline. Other places do it better; for this site to favor DocumentMode is to play to its weaknesses. Instead, this site should play to its strengths. People don't come here to use this site as a reference; they come here to participate in the conversation. This is true now; and it was true back in the days when this site was the vanguard of the XP and patterns communities. While the DocumentMode content on these subjects that was created was valuable - the ThreadMode discussion that produced the DocumentCode? content was more valuable. After all - which is more useful, the research paper, or the research itself (and the researchers' notes and journals)? While the paper is what the community at large will eventually consume; it cannot exist without the underlying work. And so it is with DocumentMode and ThreadMode - the knowledge which was documented in DocumentMode required the ThreadMode discussion to become extant.

The error that was made, in assuming that DocumentMode is all that matters, is the assumption that ThreadMode doesn't matter - and is even damaging to this site. How many useful and illuminating conversations on this site have been ReFactored? away by the WikiReductionists (on the grounds that it's clutter and nothing more), replaced with stale summaries which explain the how and the what but which have lost forever the wherefore and the why?

The reason that I come to WardsWiki - and the reason that I suspect many others come here - is precisely the conversations that we can have with our peers. For this is where the true illumination and learning is to be derived from this site. I learn things when I read a DocumentMode page that contradicts my assumptions and shows me that I might be wrong. I learn a lot more when a knowledgeable person (with a name and a reputation) tells me why and where my assumptions are wrong, and provides me the opportunity for conversation so I can learn more (and perhaps teach him/her something in the process). The value in this site is in the dynamic interplay, the extended conversation, the interaction with peers - not in pages which haven't been edited in years, and which are better summarized anyway in the literature.

So lets all raise our glasses and give a toast InPraiseOfThreadMode.

And let the flames begin. :)

-- ScottJohnson

Interesting POV, but I'm going to disagree. Problem is, there are much better venues to have ThreadMode conversations. If you want to hear the back-and-forth of opinions, why not set up PhpBB or iVisionPowerBoard? That solves all the problems we've been having lately about unjust deletions, SockPuppets, EditWars, determining who said what, determining which posts are new when there've been multiple edits, keeping threads of conversations straight, establishing reputation, and so on. Non-real-time conversational discussion is basically a solved problem: use a forum software package. Wikis are actually pretty bad at it - reading any reasonably-involved debate here is an exercise in frustration.

If Wiki loses out to CiteSeer for DocumentMode and loses out to forums for ThreadMode, why do we bother? Is it time I GaveUpOnWiki?

Personally, I always thought the big strength of Wiki was the interplay between DocumentMode and ThreadMode. It's the only system where a discussion can start out as a conversation and become a document. But I don't think ThreadMode is the endangered species here: you always have new people (or old ones) popping up with questions, quibbles, and corrections. DocumentMode takes more effort to produce, so I think the emphasis on DocumentMode is necessary to keep a balance between the two of them. -- JonathanTang

Dear ScottJohnson,

To begin with, I think you're barking up the wrong tree. DocumentMode versus ThreadMode is a false dichotomy. I think I've been on record some time ago complaining much like you do now that the obsession with DocumentMode (and the misleading ThreadModeConsideredHarmful) led wiki to gravitate towards PrematureConsensus?, or consensus at all costs. I got a lot of heat for creating CriticalSpirit and related pages, which many interpreted as an implicit accusation that community lacked critical spirit.

However, you are going all the way to the other extreme by demolishing DocumentMode by knocking down every StrawMan you can find in its defense.

I don't consider ThreadMode to be a lesser form, on the contrary, I consider thread-mode to be the preferred form for topics that are subject to research, crystalizing, topics where divergent points of views are debated, and things like that. However, DocumentMode still has to be a desirable goal to guide the discussion, and also keep the balance as Jonathan noticed. After all, just arguing for the sake of the argument has limited fun. And it's not ThreadMode that we have a problem with, it is the ThreadMess.

You claim that DocumentMode is static, or read-only, but based on what?

On the contrary, I consider DocumentMode very much dynamic because of the essence of wiki. DocumentMode is much more dynamic then ThreadMode, and certainly a hell of a lot more dynamic than ThreadMess. Why? Because DocumentMode has an essential quality: it is readable, so future readers will read and if new perspectives emerge, they will be added, the discussion will be reopened, etc (for example I wrote LawOfDemeterRevisited, long time after LawOfDemeter has been vigorously debated). On the contrary, if we live pages in a ThreadMess the net result is that they will not be read (unless by obsessive-compulsive WikiAddicts ) therefore we'll simply discourage future contributions.

You also argue against the illusion of striving for "leadership in ComputerScience", but can you name even one person who entertained such an illusion either explicitly or implicitly. In my opinion one has to be pretty delusional to entertain the idea that WardsWiki can take leadership in CS. But that doesn't mean that WardsWiki has not been and cannot be in the future a very useful resource in the SoftwareEngineering domain. The more ThreadMess you have the less useful the resource is for readers, the less attractive the idea becomes for all of us to waste our time and energy here. Why should I bother to contribute to a failed project.

Than you claim that a GentlemensAgreement cannot be reached. Well, maybe with TopMind it can't, but I can name quite a few contributors here (including you, Scott, and many more) with whom I feel very comfortable to go to the bottom of any argument in ThreadMode and then refactor without a problem and without much dispute. It's simply about assuming CriticalSpirit, picking up the responsibility of contributing with informed and learnt arguments, admitting when one is wrong or has missed something, and even if at the end of the day there are still two or more diverging views on any particular subject they can be reflected as such in DocumentMode. Even as we speak, DavidSarahHopwood is refactoring OcamlTypeSafety? problem after a long ThreadMode argument he had with me, and at the top of the page you have a DocumentMode that is pretty informative and readable. There's no problem in there. The fact that you can't keep TopMind happy unless he has it his way doesn't mean that any two reasonable persons cannot reach a finality to a discussion on wiki.

This is anti-WikiFilterist bias. DocumentMode assumes one-right-answer, and that is why my involvement often results in a mess. Usually there is more than one right answer (or at least multiple that need to be considered). I recommend some form of ControversialTopicTemplate. It allows the differing opinions to be summarized and reasonably organized. I've tried to keep thready discussions in such format over the years, but not everybody buys into ControversialTopicTemplate, creating a mix of styles (or complete lack of). To be fair, I didn't formally describe what it was until recently; trying to educate-by-example instead. --TopMind

There are so many defects in your argumentation, that I wouldn't bother the reader to shoot them all, the bottom line is this: avoid ThreadMess, conduct ThreadMode towards the goal of reaching a finality to the discussion (even if that finality will contain two divergent points of view), refactor to DocumentMode. Repeat with everyone that has a new idea on the subject. Keep ThreadMode for reference purposes in XXXDiscussion pages.

You don't need to compete with CiteSeer, nor with the JournalOfFunctionalProgramming?, however if you try to subscribe to JournalOfFunctionalProgramming? and organize a study group at work for your fellow software engineer you're likely to fail copiously. In the same time WardsWiki can be a very useful resource (and on occasion, even authoritative) for fellow software engineers.

-- CostinCozianu

CategoryWiki