Übersicht Letzte Änderungen KurzAnleitung Blog

Matching Pages:

RSS

ChangeFailure

Do you feel that the old URI cannot be kept running? If so, you chose them very badly.Cool URIs don't change

“Change Failure” is when the world changes faster than an IntelligenceProcess? can integrate the changes into it’s IntelligenceDatabase?. It’s a particular type of IntelligenceFailure.

Change Failure and Documentation

Suppose you made a wiki to document a program you were writing. You write a bit of your program, then document it as you think out ideas. You write more, document more. Write, document, write, document. Then you have some ideas, and you document still some more.

Your IntelligenceProcess? is writing documentation as you write code and think up ideas.

Your IntelligenceDatabase? is your wiki’s PageDatabase?. It collects your ideas about your program in a way that you can have information to the right info whenever you need it- much faster and better organized than actually reading through source code.

Writing documentation works, and is good.

But after a couple days, you find you’ve programmed yourself into a corner, and are going to have to make a pretty fundamental change. The time to change the code is only about 5 minutes, but now you have hours of documentation work to do. You have to check all your pages, and remove any references to the old framework, and update them to notes about the new framework.

This is a “change failure.” The world has changed faster than the ability of your IntelligenceProcess to make changes to the IntelligenceDatabase.

Particular Pertinance to Wiki

In wiki, we are vulnerable to ChangeFailure.

WikiIsDocumentBased, rather than time-based, so it has a lot of internal integration. When something is discovered that breaks the old way of thinking, the wiki is then erroneous. You can’t just change it in one single place; You need to change a lot of pages. The more fundamental the idea was to the wiki, the more pages that need to be changed. You may have to toss the entire wiki out all together.

Suppose we had wiki in the 17th century. And suppose you had a WikiPedia:Phlogiston_theory wiki. But then you discover that Phlogiston is wrong. What can you do? You toss the wiki.

A real life example was LionKimbro's wiki on WikiNodes. He was building a wiki to detail how Wiki``Nodes should work. Then AlexSchroeder pointed out that it was probably over-specifying things, and that he should keep it loose. LionKimbro agreed, but then had a lot of work to do: Vast parts of the wiki, built on the old assumptions, had to be cut out.

WikiIsDocumentBased. There is a lot of internal integration. Non-trivial changes to the wiki can have sweeping effects.

(This is also an argument for small wiki, SplitWiki.)

This is a ChangeFailure. The IntelligenceProcess? can’t integrate changes to the IntelligenceDatabase? fast enough.

Solutions

What can you do?

Another thing to do is to treat the wiki as a chronology, and view the pages as parts of a history. The older something is, the more likely there has been a ChangeFailure. It is notable that blogs are immune to ChangeFailure. Why? Because there was never a premise of integration, in the first place! See WikiOrBlog.


CategoryInformationManagement CategoryIntelligenceTheory

The above text is PrimarilyPublicDomain?.

Discussion

This one’s from WikiKM:ChangeFailure, and I’ve made some changes in translation.

Again: Alex, I hope you don’t mind me setting it to PrimarilyPublicDomain?.

People didn’t say anything about AccessFailure, so I’m assuming people don’t have problems with it.

Next up may be: WikiKM:TellingPartsOrWholes?, or WikiKM:MoreRecentMoreUseful?, or a rewrite of WikiKM:InformationKnowledgeAndWisdom.

I want to change these pages to use the SummaryBlock. They could stand some rewriting, too. If anyone wants to, feel free. (Drop me an email when you do.)

Wiki is both when elaborated, it is mostly thread mode and it is, when it is really good, document mode. At the same time. WikiisThreadandDocumentMode? at the same time - allways. It can carry a condensed common idea in a common document part of the text that was develloped out of the discussion. Blogs can’t, wiki can. Meatball has a lot of it. The base of wiki though remains the simulated direct verbal communication (here on CW not only with the friendly recommendation to UseRealNames but also to UseRealPictures?!). It is a protocol of thoughts in time. Things written earlier do not nessecarily have to be changed when a generaly change in thinking ocurrs, and in practize they are not. They are old, written before the change and they have to be understood in the context, the ambient, the shape of the antheap/termite heap (StigMergy) that exited in the very moment they were witten. Therfore the timestamp is so important. The common document, the stuff before the discussion bar has to be changed, when thinking changes, yes. But the thread mode, the protocol that tells how we achieved to develop the common document not. 85% of our communication is a just as well important protocol of these 15% that are not finely condensed yet and still scattered all over, but that are good. If these were worked out better also the work caused by a think-change would be easier to handle by a community. To my feel these things get better with more people involved. And with more time (for thinking).

Oh, what I meant was- the wiki technology is based in documents.

I didn’t mean in terms of how we actually create our pages- we talk a bunch, then we make a document, and then we talk a bunch, and then we edit the document. In that respect: Yes, conversation (thread-mode), is the “base” of our pages.

But in terms of the technology, what’s revolutionary about wiki is that it’s document based. (WikiIsDocumentBased.)

In the long run, I think we’ll see things like WikiFeatures:DocViewThreadViewSplit, (and, on it’s heels, WikiFeatures:BuiltinThreading) and that’ll all be part of Futures:TheEndOfWiki. You’ll have documents, and they’ll be cleanly segregated from our talk about them. (That would solve a lot of problems.)

Yes, what you were saying about our conversations is correct: They are not susceptable to ChangeFailure, because the timestamps and chronologies make everything clear. Our conversations are messages, not documents.

But the parts of the pages that are DocumentMode?- those are susceptable to ChangeFailure.

I think you bring up an interesting point: PeerReview? can limit the opportunity for a ChangeFailure, by nipping a problem in the butt early on. I’ll put that into the document. The community’s wisdom (WikiKM:InformationKnowledgeAndWisdom) tempers the page, so that a ChangeFailure is much less likely.

(Hmm, I perceive a need to write DocumentsVsMessages, talking about the differences between them on a particular scale.)

SummaryBlock:

PeerReview? also acts as a ColdBlanket? if applied too early, so it also slows change. Therefore it is not clear why peer review helps: Does it help fortify documents against change failures, or does it just weed out the susceptible parts? If so, are we loosing something when applying peer review to prevent change failure?

There’s a lot here.

Let’s see. We’re talking about ChangeFailure: When the world changes faster than we can keep up with it. Though, it’s not necessarily the world- it could be a realization, too.

So, an idea comes in to us (we’re playing as an Intelligence Database here,) and we have to decide what we do. We can reject it (IntegrationAndIdentity,) we can accept it. If we accept it, it could be a small change, in which case we make the small change. Or it could be a big change, in which case we have a ChangeFailure (leading to lots of IntelligenceFailure, because it takes a long time to update the IntelligenceDatabase?, by our IntelligenceSystem?.)

Already, I see two problems with the page:

  • It needs to account for realizations, not just changes in the environment. The key thing is that it’s a new something (internal realization, change in the external world) requiring many changes.
  • The IntelligenceFailure happens because it takes a long time to integrate the many changes into the IntelligenceDatabase?. (It could be stated more directly, in the DocumentMode? text above.)

Still haven’t gotten to what you are saying.

:“PeerReview? also acts as a ColdBlanket? if applied too early, so it also slows change.” Therefore it is not clear why peer review helps: Does it help fortify documents against change failures, or does it just weed out the susceptible parts? If so, are we losing something when applying peer review to prevent change failure?

Hmm… I still don’t get it.

ColdBlanket? doesn’t need to be shivering; It can be “Hm, are you sure?”

The purpose is to slow the change, or to prevent a change entirely. And that may be exactly what we want. Because it could be a harmful, or erroneous change. If I write OnePlueOneEqualsThree?- “Great new theory,” I would hope that people would argue, “But, consider OnePlusOneEqualsTwo?- how do you refute the great X plus X making XX, because you put them side by side, and XXisTwo??” Thus, preventing the change. So, the slowing of change is good in that respect.

We are saved from a ChangeFailure: The realization that our entire wiki on (say) mathematics is completely bogus, because it’s founded on OnePlusOneEqualsThree?. The PeerReview? saved us from a ChangeFailure.

I don’t think we’re losing anything.

It may happen that we cut off perfectly good ideas, because it’s actually our base assumptions that are wrong.

It’s recognition that we can be wrong in our fundamental, base assumptions, that makes us want a RightOfMentalPassage?, and practice InternetWide? RadicalInclusiveness. We may not want an idea integrated into our own IntelligenceDatabase?, by IntegrationAndIdentity, but we’d like it to be integrated into some IntelligenceDatabase? out there. That way, if we’re wrong, the idea has safely germinated in another place.

OH! But: While I was writing the SummaryBlock, I think I learned what you meant:

What if PeerReview? has the effect of freezing our thoughts? What if- what if we become Conservative. What if we fall in love with our ideas, with the PageDatabase?, and then reject big changes, just because it would entail messing up our beautiful pages. What if the PageDatabase? perverts our reasoning, and we reject ChangeFailure just to avoid changing all the pages. We might not do it on a conscious level, but we may well do it on a subconscious level.

I know that- when I worked on my goliath notebooks system- that I experienced just this problem. I called it “mental freezing,” and I even wrote about it.

Is that what you are talking about? Should we make a new page, MentalFreezing?, to talk about what happens when an IntelligenceDatabase? overwealms our mind, and keeps us engaged in mapping out more territory by an old paradigm, rather than jumping ship and moving to the new understanding right away?

In that respect, PeerReview? as a system of avoiding ChangeFailure: It seems to me to go back to IntegrationAndIdentity. The conclusions are still the same:

  • PeerReview? can help you from making dumb mistakes, and keep the IntelligenceDatabase? clean. This prevents ChangeFailure.
  • PeerReview? can stifle important and necessary changes. In this case, it causes a ChangeFailure, since eventually, the community is going to realize that it’s out of touch with reality. (Eventually.)

I believe that, all things being equal, it’s better to be checked by a community, than not to.

The system goes like this: You find a community that shares similar values and goals with you, and understands things more or less the same way. You work with that community to flesh understandings out, and discipline yourselves to your values and goals. The community helps keep you in check, and you help keep the community in check.

But one day, you discover something that changes your thinking, in a way that you can’t reverse from. Perhaps you see a bigger picture. Perhaps you see one of those Deeper Synthesis, described in IntegrationAndIdentity, and you are not able to persuade the rest of the members of the IntelligenceSystem? that the deeper vision is worth integrating into the IntelligenceDatabase?.

In this case, you go somewhere else, where the people share the similar values and goals with you, and understand things more or less the same way. Then you work with them in their efforts.

Until, once again, you move on.

I believe we need a name for these ideas. This is a pretty fundamental notion, to what I believe.

(However, you always remember where you came from, and Love your friends, regardless of mental disagreements, and so much stuff. It’s all just a bunch of message passing, after all. Love is something more important.)

SummaryBlock:

  • Recap: Ideas come in. Accept or Reject. (IntegrationAndIdentity.) If accept, Small change (ok) or Big change (ChangeFailure.)
  • Problems with page. (A Digression.)
    • Needs to count for internal realizations, not just changes in external conditions.
    • Directly say: ChangeFailure is an IntelligenceFailure because it takes a long time to integrate the many changes into the IntelligenceDatabase?.
  • Alex’s question:
    • Are we losing something by using PeerReview? as a ColdBlanket? to avoid ChangeFailure?
  • The purpose of the ColdBlanket? is to slow change.
    • It keeps our IntelligenceDatabase? from corrupting, and eventual ChangeFailure.
  • Trying to cut off bad ideas, we may accidentally cut off good ideas.
  • Trying to preserve good ideas, we may accidentally preserve bad ideas.
    • Since we don’t know, ultimately, good from bad ideas:
    • Preserve RightOfMentalPassage?, practice InternetWideRadicalInclusiveness?.
    • IntegrationAndIdentity makes us need to protect our database.
    • Humility makes us need to make sure people find a way to germinate their idea.
  • MentalFreezing? can occur, if we fall in love with our IntelligenceDatabase?.
    • We may become conservative, and PeerReview? good ideas out of existance.
    • But, it’s good to be checked by the community.
    • (Thus, it is a necessary danger.)
  • IdeaSamsara? (WikiPedia:Samsara - “Sam: Together, Sara: it Flows”)
    • Find people who share your values and goals, and work with them.
    • Eventually, move on, when you realize something that everyone else can’t or won’t.

(Sorry for the rambling; Hope you could follow it..! I’m ambulating through this territory, myself, trying to figure it out.)

Did any of that make sense? I ask, because I look at the clock, and it says it’s midnight, and I know from experience I don’t make a heck of a lot of sence around midnight.

LionKimbro

This is exactly what I’m talking about. I like “What if we become conservative? What if we fall in love with our ideas, with the PageDatabase?, and then reject big changes, just because it would entail messing up our beautiful pages?”

Oh, okay.

Well, in that case, we have us a MentalFreeze?, which leads to a massive ChangeFailure.

;)

But, you know… There are all sorts of places in the world where there is a MentalFreeze?, that sort of persists, on a social level. There are, you know- just to pick something- all these Ayn Rand-ites, who do not see the flaws in their system. One day, they will. But, they will not be able to convince all their buddies. People move in, and out, of that particular system of thinking. (I do think these people should have a wiki, of their own.)

LionKimbro

Fans of the Vingean Singularity tend to call this “future shock”. Try reading Yudowski’s work, such as [1]. The point is that as we approach the mythic Singularity, the rate of technological change speeds up to the point where nobody can cope, at least using present techniques. If the Singularity hypothesis is correct, then ChangeFailure is a vitally important issue.

Yes, I am aware of this idea. I participate on the SL4 wiki every now and then, and consider myself “at” Shock Level 4, by their strange terminology. (I don’t know- it strikes me as sort of goofy. It reminds me of Scientology “OT” and de Bono’s “Categories 1-9” scheme for people. I’m not too fond of these systems, but regardless- I’m flashing my colors.)

My feeling is that most people have zero idea what’s coming straight at them, and how little influence anyone has over it. That’s what we have movies for, though, I guess. Minority Report, Terminator 3, I, Robot- they do a good job of explaining things, I guess.

I’m not sure I agree with the overall point of this page, at least in terms of comparing wiki to actual real alternatives. If you make a big mental change, then until you re-document the entire scope of that change, other people have to guess which bits of the archive are still accurate.

Hmm, let me try to get more clear. Let’s compare managing change between a wiki and a blog. Let’s assume we’re talking about some chunk of thought that takes a fair amount of written language to express. Let’s say it would take up the equivalent of 5 mid-sized wiki pages. And the previous version of that thought has already been fully documented. In this wiki that might involve 5-10 (or many more) different pages (since you break things up logically rather than just on the basis of pagesize - DecomposeIntoLogicalChunks?). On the blog you may have 1-5 (or many more) documents covering that thought (on the assumption that blogs don’t encourage the DecomposeIntoLogicalChunks? pattern, so there’s a greater risk you end up with a big reference page more logically similar to an old Word document).

So now you have this change, which is in your head, and it will take time to document. Maybe more time than you have right now.

I would argue it’s ‘harder/slower’ to incrementally update your OneBigBlogPage? than your 10 wiki pages.

Hmm, I guess I’ll need to come back later and try to back up that claim. But I hope I’ve defined a bit of context that might lead others down that same train of thought.

Sorry I took so long to respond! Somehow, I just wasn’t paying attention to RecentChanges, and didn’t see it.

Anyways: I think we’re just misunderstanding each other.

When you participate in a wiki, and there’s a major realization, that shatters the value of say- 20 pages- then you feel some obligation to go back and revise those pages.

That’s because the wiki represents something of a living endorsement.

But in a blog, if you think something was wrong in the past- you don’t go back over your whole corpus, and edit it out. Doesn’t matter how long the corpus is, or how many pages your old idea was in: You just don’t edit old pages.

A demonstration:

If your blog says…

http://www.bigreds.com/TURTLE2.gif
“The best teenage mutant ninja turtle is clearly Rafael. He was the coolest. I mean, he’s got the Sai; what weapon’s cooler than that? – LionKimbro, age 9”

… then there’s no reason to edit it out. I mean, you were 9 years old.

But if it’s your wiki, and it still has a page WhichTeenageMutantNinjaTurtleIsCoolest, well- that’s probably something you want to edit out. ChangeFailure.

Just some more recent thinking:

The brain seems to have an amazing ability to handle change. If you spend your life for 10 years by one philosophy, and then have a major change in philosophy, the brain does some pretty astonishing things, compared to a computer, which is incredibly hard to change. In computers, changes have to be carefully prepared for and applied, or everything breaks.

So, how is this possible? I can think of two possible explanations.

  • Possibility 1: The human brain is capable of maintaining multiple explanations A at once. If one turns out to be false, it switches to explanation B, which it is already familiar with.
  • Possibility 2: The human brain is incredibly dynamic, and we think most things through at run-time. A “big change” in philosophy, then, is not really a big, system-wide change. Rather, it is really just a little change, and we can handle it, because we’ve always been extrapolating a great deal off of a little itty bitty bit- not relying on hard-coded entrenched systems.

What would it be like if Possibility 1 were true in computers? It would mean that you programmed computers by multiple explanations at once. When the computer wanted to know how to do something, it would try a few explanations. At the same time. If one fails, … eh, use the other one.

What if Possibility 2 were true in computers? Then they would be super-super-runtime reasoning programs, relying on a very few “facts” at a given time. You change a small fact, and it’s okay, because it’s always looking 2,000 chess moves into the future anyways.

All this said: I am no expert here. I have no idea what I’m talking about. I’ve not once taken a course in AI. These are just some thoughts I found interesting, related to ChangeFailure.

Yeah, I agree (with both possibilities). Of course, this is purely speculative on my part as well.

"The 'always-beta' culture is affecting more than just journalism" by Elias Bizannes 2009 talks about “Soft knowledge versus hard knowledge”. He says that the vast majority of knowledge is only in the “keep it in your head” form.

And he argues that new Web 2.0 tools give us a third option between

  • write it down, spend a lot of time and effort making sure it’s accurate, and then publish – perhaps too late to do any good; vs.
  • don’t write it down, keep it in your head – but then that information is unavailable when you’re not there.

What if Possibility 2 were true in computers? Then they would be super-super-runtime reasoning programs, relying on a very few “facts” at a given time. You change a small fact, and it’s okay, because it’s always looking 2,000 chess moves into the future anyways.

Check out the work of JohnHolland?, particularly his book Emergence. It is possible to run computation on EvolutionaryAlgorithms? that do this “think-ahead” that is described in LionKimbro’s comment above. Some of this was also covered by StephenWolfram? in “A New Kind Of Science”.

An interesting idea is that a computer could do a combination of 1 and 2. A program “learns” by playing out many scenarios. Yet, a program may also “learn” in a NeuralNet? way, by “JigSawComputing?” (http://ieeexplore.ieee.org/xpl/freeabs_all.jsp?arnumber=166400) which is how the brain actually accomplishes computing. Our brains do less of the playing out scenarios, but a computer could do both all the time, potentially.funny quotes funny jokes divorce attorney hilarious quotes dental implants

Umleitung auf eine externe Seite definieren: JigSawComputing OneBigBlogPage PageDatabase EvolutionaryAlgorithms JohnHolland IntelligenceDatabase MoreRecentMoreUseful DocumentMode IntelligenceProcess MentalFreezing OnePlueOneEqualsThree IdeaSamsara RightOfMentalPassage StephenWolfram XXisTwo NeuralNet TellingPartsOrWholes InternetWide ColdBlanket MentalFreeze OnePlusOneEqualsThree IntelligenceSystem OnePlusOneEqualsTwo PeerReview NoWriting WikiisThreadandDocumentMode PrimarilyPublicDomain UseRealPictures DecomposeIntoLogicalChunks InternetWideRadicalInclusiveness

Sprachen: