Go to Making Light's front page.
Forward to next post: Robert Legault
Subscribe (via RSS) to this post's comment thread. (What does this mean? Here's a quick introduction.)
Somewhere in the war between enthusiasm and cynicism, the content of Patrick’s notes on the O’Reilly Tools for Change for Publishing conference went undiscussed. And I’m sorry about that, because some of them really got my attention. They looked neat. I wanted to hear more.
Then, less than a week later, I was sitting in a restaurant in the Alps, nursing a bruised tendon in one knee while one of the smartest people I know dozed beside me. And suddenly two of those one-line comments tied together a mass of notions I’d collected from a dozen disparate sources. It was pure high country cascade, and I’d like to pick apart some of the ideas from it here on Making Light.
‘Content is king! Content is king!’ No. Nor is ‘context.’ Contact is king.
Since when is the Internet a monarchy?
Content is important, since the internet is basically an information delivery system. But we’re social animals, so we bring community into everything we do. Look at the stories of the desert fathers; these men were hermits, and still their lessons are about community. We’re the same today, even when we’re alone at our keyboards.
Really good communities tend to grow around valuable or interesting content. Google and StumbleUpon pull new readers in. If the content keeps moving and stays good, a certain proportion stay around and chat, get to know each other, hang out. The growing community at BoingBoing is a recent example.
Contrariwise, really good content is about community. Writers are motivated by readers; creation implies audience. And a good community can generate even better content than an individual, if conversation and dissent are used to refine the ideas under discussion.
The two are one, like hands joined together, like the end and the way.
Curating the conversation will be a whole new kind of editorial job.
Communities that generate content generally do it in conversation (not always; my old community didn’t). That conversation needs to be gardened and nurtured while it’s in progress — and that’s a whole ‘nother ball of fishy kettles — but it also needs care afterwards.
Online conversations can strain even Sturgeon’s Law. Some of them are astonishingly crap. And even — or especially — the good ones wander all over the place, cross-fertilize, merge and separate. Finding anything again is a matter of luck and memory. Finding it the first time is more about miracles and Google-fu.
That’s not going to be good enough in the future. More and more new ideas are first proposed online. Without a record of how they were developed, without the initial arguments against and reactions to them, they lose a significant part of their meaning. (For instance, only without context can Wikipedia seriously describe disemvoweling as “a technique by forum moderators to censor unwanted posting”.) Valuable historical information is being lost.
One could manually index conversations, of course, but that’s enormously time-consuming. Kelly McCullough’s excellent work indexing Making Light and Electrolite is the product of great effort as well as love. To index every conversation like that is way over the event horizon of practicality.
So what should be done?
If I may go all Charlie Stross for a moment (by which I mean buzzword-rich and science fictional), we need some way of automatically tagging and flagging conversations.
Tagging can indicate the content of a discussion. Word-frequency analysis might be useful here, perhaps in a moving average over multiple comments to pick up threaded discussions. There is a tension between using a controlled vocabulary and a free one; one is more searchable, the other more adaptable. Perhaps some form of evolving or semi-controlled vocabulary? Bayesian algorithms already figure out which emails are spam and which are ham; could they figure out which tags are applicable and which have been superseded?
But conversations should also be flagged for quality. A good conversation is idea-rich; a poor one reiterates old notions and beats them to death. Text mining and search systems should prioritise the former over the latter. And that’s yet another question: what are the markers of a good online conversation? I bet that’s harder to detect automatically than trolling, even, but what a treasure for searchability if we could!
At this point, my companion woke from his sleep and listened to my excited handwaving exposition. He was unconvinced.
What say you?
Nota bene: constructive criticism of goals, means and methods is keenly sought. Rants might be better left at the door, unless they are funny or in verse.
It could be argued that the very lack of history returns us in some ways to myth.
Or, if not to myth, to genaealogy.
...what are the markers of a good online conversation?
Being me, I'd look for indicators of bad conversations, ignore tham, and whatever is left over ought be of superior quality. Looking for new ideas and content sounds hard, because we don't know what it will be. Instead, filter out all the things we've heard already too many times.
If I may go all Charlie Stross for a moment...
Not going for the shaved head and bag of gadgets?
Back when I was job-hunting while hammering the last few nails in my linguistics PhD, I interviewed with a company that was developing cognitive-based tools for determining what they called "the aboutness" of electronic documents. Not simply a word-frequency based guess at topic, but an interpretation of attitude, genre, etc. I never did get a chance to find out more about the approach since I wasn't called back for a second interview. (I've always wondered if that was because I commented on how useful their project appeared to be for illegal wholesale governmental spying on electronic communications. But hey, maybe I'm just paranoid.)
What makes a good conversation anywhere is the willingness of people both to speak and to listen, and to do more of the latter than the former.
Also, since abi wants verse:
your chances come and go like a spring breeze
above the tulips maples still are bare
but all the city seems to be aware
that something's brewing in the mysteries
nature may hide some trick in her chemise
that the best gardener would not think she'd dare
and then send signals out in simplest clear
when we most think to sit and take our ease
on edge of spring we wait as on each night
the stars reveal another sort of chance
and we are given leave to ask for rest
not knowing yet what we may get as right
nor what our steps are in the coming dance
but hoping that each change is for the best
I have a blog post from awhile back where I called out what I saw as the 4C's of the new social media: Content, Community (or Conversation), Consumer and Commerce.
Quote: "If Content Is King, then Community is Queen. A fickle, hard to please queen at best. Easy to plan and support with technology, extraordinarily difficult to nurture and grow."
Are there two points here? 1:Contact is king (or at least coequal with content) and 2:There's a need for ham-recognition software.
Perhaps there's room for a sideways corollary: there's a need for contact-recognition software. If 'ham' is content, then 'feast' could be called the enjoyment of ham.
There's only so many hours in a day, so it's advisable to spend a certain amount of it keeping up with existing community - 'feasting' - and a certain amount fishing for new content and/or community - 'feast-seeking'.
A feast-seeking agent would be one that incorporates ham-recognition while bearing in mind the seeker's tastes and willingness to experiment ("My seeker likes to talk about A, so I'll suggest A' to them but not B.")
I was sitting in a restaurant in the Alps, nursing a bruised tendon in one knee
...and nursing a beer in the other knee?
"Have another knight cap, dear."
Isn't the community itself a content?
No, I'm not sure what that means. It just popped in my heard.
Abi, maybe, instead of relying on algorithmic processing of text ab intio, we could start with some framework?
Back to community markers: most things that TNH, or Charlie Stross, or Scalzi, or Gaiman write are worth reading - have quality - say 90% quality. They are very valuable due to their quantity, too: instead of searching everything, everywhere, for relevance (the bind google algorithm way), one merely need wait for them to tackle the topic. Given an inadequate, but still large amount of text, and an inadequate assemblage of great writers, the chances of finding a relevant "answer" are pretty good.
Our ancient tribal skills of recognizing really good storytellers (and truthful ones) is a good enough algorithm, I think.
Ahem. Well, I meant to be more insightful. This isn't working out well. Let me try with numbers:
Scenario: you want to understand something. You have access to:
20 people* who cover 80% of everything with 70% more text than most people can muster.
70 people** who cover another 10% of the world's knowledge, but do it with about 20% more text than the average person
250-500 people*** who cover the remaining 8% of the worlds knowledge, and do it with about the same amount of text as an average person.
That's not that many blogs to search through, or books to read, but a primitive sorting algorithm might be to start skimming group one right away.
Note the missing last 2% of the world's knowledge, which is the property of the dead.
I lost my point. I'll try again tomorrow.
*e.g. TNH
**e.g. Joel Spolsky of Joel on Software
***e.g. Ashcraft on Kotaku's Night Notes
I'm glad I checked the bottom of my browser window to see just where that Google link led. I'm not sure I want to know the results of that search.
We have almost no writings of the Greek poet Archilochus, who was revered by his contemporaries, but we do famously know one line he wrote:
The fox knows many things, the hedgehog knows one - one big one.
Do you want your conversations to tell you what the foxes think, or to get the occasional hedgehogs to uncurl a little and tell you what they think? 'Cause, you know, they've got to feel ready to uncurl before they can tell you.
(Funny, much of my recent work-life has involved algorithmic social networking tweaks....)
Precisely because context is so important, I think flagging is a very tall order, a function of the text, its referents, and the reader and theirs [sic].
Example: I found American-style Libertarian arguments fascinating the first time I heard them. It was new information and a new way of looking at the world, and so I probably thought the writing was better than I would now (this, incidentally, is why pornography succeeds in new and flawed mædia---interest level substitutes for quality of presentation). Now they feel like an algorithm I understood awhile back; my judgement of the "interestingness" and (more significantly and sadly) overall level of quality of any such would now be much lower---pretty much anyone not as good as David Friedman doesn't seem worth my time.
Of course, you could keep a couple of readers on ice at the National Bureau of Standards....
There is a new technology being developed that can perform automated tagging, from a company called SearchPhysics. It detects and exploits patterns inherent in any language (and in many natural processes, such as galactic clustering) to extract knowledge about the subject material presented. The neat thing is, it's language-independent, and requires no training.
Automatic mechanisms for classifying or indexing poorly-understood domains strike me as problematic. What I usually ask of automation in the beginning of any major undertaking is assisting the work of an expert* rather than performing all the work. So what I'd like to see is tools to make it quicker and easier for a person to identify and classify conversations, and to detect changes in the characteristics of a conversation, so it can be reclassified as necessary.
Slashdot, Digg, and Stumbleupon are different ways of finding interesting conversations or creating them around interesting content using social networking with simple rules based on human interaction. More complicated and subtle (and varying) rules could be used with some automatic tools to help. What's interesting to me is that the evolution of these kinds of sites over the last few years has been through a continuous accretion of layers of meta-organization and aggregation, with technology mostly used to connect the layers or aggregate the components rather than to detect or help create them. I'm not sure if this indicates that we don't know what tools would be useful, that such tools are very hard to create, or if it's just quicker to figure out how to get people to perform such tasks, now that getting them to communicate is easy.
* or someone who's trying to become an expert
Distribution rules--without, there's no information communicated.
The best whatever in the world is utterly useless if it's in the wrong place or has the wrong people for operating it or gets operated improperly for a purpose it's the wrong tool for.
Distribution, when it comes to content availability, trumps quality--consider Fux Noose et al making Gresham's Law look optimist ic. If someone's never been exposed to quality goods, or never taught what and why about "quality," or how to think critically, or is obdurate about dismissing the scientific method in favor of uncritical credo... that person is probably going to not be able to distinguish crap from quality and won't care about distinguishing crap from quality.
Regarding community and quality, one size almost never fits much less flatters all. (Flatters might not be the word I want, the word I want denotes adornment that improves how someone appears and puts them in a positive light, highlights them positively, etc.) There are things as simple as, "I am a short, small woman. Those becoming ubiquitous electronic signing things in stores, tend to be at heights which are difficult for me to read the text displaying on them, and even more obnoxious as intended for being "signed." Someone I do NOT think that the people who designed those setups, included, much less bothered to consider the existence of and usability to, of short people. They are noxious, aggravating, irritating, and rude as regards my opinion of both them, and the people who implemented them at heights inimical to me. Someone a
half-meter taller than I is likely to have a different opinion.
That is an example of a physical difference, that makes a community difference in effect. The store usability regarding those pieces of equipment, applies much better to Tall People than Short People.
Regarding content, there are lots of individual differences of perception, experience, background, physiology, etc., that affect how different people interpret the same words. Some groups have encoded messages and terminology, particularly that applies for things allegorical, with the equivalent of a mental dictionary of substitutions involves, shared across "community" or shared to some degree across communities.
Texts get redacted over time, and different groups conflict as to what the "correct" text is and what the correct interpretation/interpretations are. It gets particularly vicious where there are translations involved and the world of the author(s) recede away in time and culture.
But, the same words assembled together, don't denote the same things to different people. The phrase that one person admires and finds of surpassing brillance to one person, is the purple prose, or boring, or devoid of life, passage to another.
One bottom line on at least one axis, reflects what the work evokes, or fails to evoke, in the individual reader. The reader brings the reader's conventions and experience and taste and worldview and history and values and expectations to the material--and those are things than can and usually do change over time. The words and the order they're in stay the same, the person reading them changes.
Communications theory examines all of the sender, the transmission media and channel, and the receiver--anywhere in those the message can be distorted, misencoded, misinterpreted, mangled, mishandled, dropped, decoded with an incorrect key....
"Quality" is a judgment issue, and the criteria and how they're applied mattered.
Lots of people never look or even want to look at their assumptions about what is good, what is bad, what is "quality," and why they have the values and beliefs and criteria they have--if they are even aware of most of their values, beliets, and criteria.
"Why do you like this?" or "Why do you value this?" isn't necessarily a question that most someones ask themselves, and without knowing accurate answers to those and other questions, the issue of "quality" becomes an uncalibrated conflict of belief and worldview.
"I liked this because it reminds me of the place I grew up," is a valid reason for liking something, but it's not necessarily a criterion for someone being quality literature! "I despise his prose style," explains why someone might put a novel on a Hugo ballot below No Award, but enough other people regarded the book (or perhaps the author...) highly enough nominate the work for a Hugo.
What I'm proposing above, is that context matters. "Give me a lever to push with and I will move the Earth," involves there having to be some fulcrum available for the lever, a person to push on the lever, and of course the Earth for the person to push the Earth with. The fulcrum usually goes unremarked and left out as requirement.
With tangibles involving levers and such, success, failure, and metric are much clearer than quality of a literary endeavor.
But getting back to distribution again--how is someone to attempt to measure quality, without criteria and standards such to measure against, for comparison and contrast... distribution denotes availabiilty (hopefully) of the work, and people that it gets to, being able to do their own metrics, however qualitative/quantitative they be ("This book has a lead character from a place I detest as protagonist, so it's a bad book!" "The author wrote this book in iambic pentameter, I love iambic pentameter, and it is the most wonderful book because what I care about most is a regular rhythm and meter!" "This book has 800 pages. I LOVE big thick books, this book is wonderful!" "This book is an 800 page big fat fantasy novel. BFF novels are evil and despicable and I hate them so this book utterl stinks!!!" )
I think Paula and Bruce are both getting at some very interesting angles on the idea, both usefully avoiding the "solution in search of a problem" or "technology in search of an application" that I see so much of in software.
Paula rightly identifies one of the core questions as "quality in terms of what values?" and Bruce notes that genuinely useful software tends to focus on figuring out important work that people do and finding ways to assist them in the repetitive grunt-work parts of it, rather than trying to replace the people. (Word processing software has changed the world. Search engines have changed the world. All attempts at developing software to write fiction or non-fiction have gone nowhere.)
How do you beat Sturgeon's Law? Communities. A community of like-minded folks, a bunch of people with the same (or at least similar) interests as you are always going to be the best predictors of what you’re going to like. This is why good content generates communities, and good communities generate content.* It is the function of one to find the other. I think this principle is widely understood on a subconscious level, but widely misapplied.
Take community ratings systems, like Digg or on Daily Kos. I've always been really disappointed with the way these systems work out. The critique you always hear about them is that they inevitably become a self-reinforcing cycle, where the highest-rated blogs/whatever are most widely read and so they get the highest ratings, and low-rated ones stay unread and get low ratings. That isn’t what bothers me. What bothers me is that they don’t work.
It is commonly accepted that general popularity is a good predictor of quality. From the individual’s perspective, that simply isn’t true. Trying to guess what people will like by applying statistical averages is like trying to guess people’s height with a bell curve: you’ll be right more often then if you guess, but you’ll still be wrong more often than right. It’s simply not a refined enough technique to surmount Sturgeon’s Law**—50% crap is still too much crap. Wide appeal simply isn't wide enough. People want things that appeal to them personally, and for every person, it’s going to be different.
Community ratings systems fail because they focus on the “group” part of communities, rather than the “people” part. It’s people, specific people, that are key. People join communities because the communities are made up of people more like them than the average population. That’s why they trust them. Communities are communities specifically because that can’t be replaced with a random sample. Even within communities, particular people matter. When I want to find some new music, I don’t take a poll of all my friends. I ask my friend who I know has good (=complementary to my own) taste in music. The input of thousands of people is less useful than the input of one person who you trust.
The best tech fixes, especially when it comes to social stuff, are ones that leverage existing strategies, rather than trying to build from scratch. So how can we take this basic, fundamentally personality-based crap-sorting mechanism and turn it into something that can deal with something as huge as the Internet?
Imagine a Digg that’s designed to track individuals’ tastes. Instead of creating a single front-page list, every list is personal: as each user browses around, they tag things they like, and people they like. If a comment or a post strikes them as particularly intelligent, that alias/person (independent of site), along with the comment, is added to a list of People I Think Are Neat. If a Person I Think Is Neat posts to a blog, or some of them show up on a comment thread, the user gets notified. If the user thinks someone is really extra neat, they can go a browse that person’s list of neat things, and see if any of them appeal. It’s a system designed to privilege personal connections over group identities, individuals over sites. Each individual can create a hand-tailored community of their own, reflecting as precisely as possible their own quirks and interests.
*If good content produces good communities, and vice versa, can that teach us anything about why communities fail? Do they fail because their content sucks? (Example: Arguably, the most of the content that Wikipedia produces is Wikipedia rules lawyering. This is clearly worthless content in any other context. Is that why the Wikipedia community is so toxic?)
**Let me propose a corollary to Sturgeon’s Law: Everyone can agree the 90% of everything is crap, but no one will agree which 90% that is.
From a different perspective -- and I mean these to be honest questions, not rant, even if they have a touch of the Devil's advocate to them -- what's really new here?
People have been having conversations since the birth of language. Nearly all of the contents of those questions are lost to us; nearly all of the things that were developed in those conversations are things that we only know of when they touched archivable media. The parts of the OED discussing pre-15thish-century language are a veritable mass of this; things that obviously had been in use hundreds of years before the first written example. That conversations get lost is not new.
It's not even that more is being generated; if I were talking instead of online, I'd be generating lots and lots of spoken words, which would be heard by one or two or three other people, and then gone. But more per person is being audible for much longer; here, dozens, maybe a hundred people will read this, some half-dozen crawlers will archive it, the site will stay up for ages, and some of those archives will likely persist until God-knows-when.
What is new, then, is that these conversations are occurring in a way that (a) they can be listened to by many people over a period of time, and (b) they can potentially be saved without much effort. In meatspace, I can participate in two or three conversations in an evening. Online, I can easily read two dozen. And, if the conversations are interesting, I can save them in various ways. I have four dozen Livejournal pages open to read when I have more time; a large pile of bookmarked posts; some hundreds of megabytes of things saved to disk; and almost uncountable numbers of Usenet posts that I forwarded to my email account because I liked them.
And I almost never read any of these more than once.
So, then, my question: What is it, other than the fact that -- like Everest -- the possibilty of archiving these conversations is there, that makes it worth saving them?
We have this idea that anything that can be preserved must be preserved. Could it be that this idea is simply an archaism in the modern copyable world; a quaint leftover from the time when the text a person could accumulate would fit in a few bookshelves?
Beyond that, do conversations preserve their worth beyond the interaction of those participating? I have read arguments about -- for example -- the length of functions in programs, and whether or not there are legitimate reasons for ones longer than 20 lines. And I know the answer to that, and the reasons; there is no worth in rereading that. Having the argument once -- and participating in it, even if only to form the reply in my head that I would make if I chose to be bothered -- was valuable. But, as dead archived matter without the opportunity to reply, there's little point in reading ill-formed rants and immediate reactions in the comments. So why archive it?
Perhaps the right solution is the way things have always been; that conversations should only be archived in the effects they have on those participating, and in whatever rare gems those participants are willing to transcribe longhand.
heresiarch @ 18: That system is, to a small extent (and with a bit of an inversion), why I find that Livejournal works so well for meeting people. I read journals of people I find nifty. And I read the comments on those journals, which are generally by people specifically who also find that person nifty. Reading the comments gives me an idea of what they're like, and sometimes I add them to my reading list too, and the process propagates -- and I have what I thought was a small reading list that has probably a hundred people on it, most of whom I suspect with opportunity and time I'd be pretty good friends with.
I've also found Usenet groups by DejaNews-searching on the names of people I liked, back in the day.... (And, yes, that was back in the day when it was DejaNews!)
I wonder if there's a historian out there wondering what the early days of the internet were like. I managed to archive almost everything that was ever posted to EMUSIC-L, an electronic music mailing list that ran from 1989ish to 2000; it was a fascinating place at the time, mostly because there really wasn't anywhere else to go if you didn't have Usenet - anything from respected academics to casual hobbyists, all chatting animatedly.
. I wish I'd known what I know now about cultivating a community from reading Making Light; we'd probably still be running, in some form or another. And I would have had the distinct pleasure of disemvoweling one particular creep, who I think managed almost single-handedly to destroy the community, between his disruptive behavior, and what I now think must have been sockpuppetting or behind-the-scenes rabble-rousing.
My guess is that his/her posts would now be about three characters long each, after removal of line-noise and excessive whitespace. And every vowel in sight.
I'd never previously thought about it quite so much as a historical document,; I guess it really was worth it.
Nomie, @11:
Actually, that google search mostly leads here. There's nothing particularly NSFW (except the word itself) in the first page of search results, though there was one page that left a nasty taste in my mouth.
*tries the Google Image search* Maybe I have safe search on, but it's hard to tell when my search engine is speaking Dutch. Yep, it was on. There's one midly dodgy picture, but essentially it's safer to image search for dinosaur sodomy than it is for twister.
Brooks Moses @ 20: Yeah, I think that lj definitely works along similar lines. The big difference, I think, is that lj is a people-driven process for finding people, where what I have in mind is a people-driven process for finding neat ideas and conversations.
After "content is king" there was a period when the buzzword was "content is dead". I was tempted to put up some pages filled with "alk;jscfnkweio asiodfnklcx welhhflkqoiqjklc" to celebrate the death of content.
#18, heresiarch made me think about del.icio.us and why it is popular.
I don't pay much attention to Digg or Daily Kos. Honestly, I don't pay much attention to del.icio.us. But I think that one of the things that made del.icio.us so popular is that you can find someone with similar tastes in a very specific item, and follow their links tagged with that topic.
So if I had multiple food allergies and were looking for recipes for baked goods, I could find the Gluten-free Goddess' del.icio.us account*, find her tag "baking" and sub to an RSS of that very specific thing. I'm not trusting a community to bring me things of interest, I'm following a person.
The problem is finding the people whose taste in food/music/conversation matches what you're looking for. Or is that just restating everything that's been said before? Maybe all I'm contributing is the realization that if Digg and Daily Kos don't suit, del.icio.us is an alternative. There are others as well - furl is one, but I think del.icio.us' popularity makes it likely to be the most useful.
The ability to follow individual tags/conversation threads with RSS seems to me to be an incredibly valuable contributor to finding the stuff that's interesting to an individual, but I am definitely still in mad love with RSS as a concept** and could be wearing rose-colored glasses on that point.
I love the corollary to Sturgeon's Law. Well done.
*I don't know that she's got one, I'm making this up as an example.
**It's a bit embarrassing, that adoration. I've been playing with RSS and readers for two years now, so I'm hardly a newbie., but I keep thinking it is utterly awesome, especially when I find places I can sub to specific tags so my signal-to-noise is very high.
Clay Shirky's article Against Well-designed Reputation Systems is essential reading. He makes a pretty compelling case that successfully implementing reputation/labeling/flagging systems can only occur after the community's culture has begun to develop.
His premise:
"...the need for some user-harnessed reputation or ranking system can be regarded as a foregone conclusion, and that these systems should be carefully planned so that tragedy of the commons problems can be avoided from launch. I believe that this conclusion is wrong, and that where it is acted on, its effects are likely to be at least harmful, if not fatal, to the service adopting them."
His conclusion:
"I do however mean to say that the central design challenge of user governance — self-correcting systems that do not raise crushing participation burdens on the users or crushing policing barriers on the hosts — are so hard to design in advance that, provided you have the system primitives right, the Boyd Strategy of OODA — Orient, Observe, Decide, Act — will be superior to any amount of advance design work."
The subtle point here is that we are a long way off from being able to develop standardized meta systems for social media content. The best we can do is implement community specific systems.
Feel free to conduct this thought experiment: A book publisher wants to create a mega community site around the community of SF authors, readers and publications. How is it possible to determine in advance what kind of reputation, flagging and tagging systems you can successfully shove at a bunch of contrary, pig headed, fiercely independent, privacy loving, anti-group thinking (ok ok you get the point) bunch of fans and authors? Yeah, good luck with that!
I wonder if it matters who the curator is? Is it the owner, organizing discussions "somehow" with [insert Nifty Software here]? Or is it the reader, collecting information of personal interest? The algorithms and constraints that a site owner might choose might not be useful for others (Heresiarch's Corollary!).
I can also imagine the owner of a site thinking, hey, I really enjoy all the discussions, but I have neither the time nor the inclination to organize this for posterity. On the other hand, a reader might very well want at least some of the information. Something like this actually came up very recently in a comment in the Bérubé thread (also, comments 36 and 39), although the emphasis was on the posts rather than the discussions about them.
So, while we are debating the utility of tagging, community ranking, and other ways to organize a conversation, I had a thought:
Any given post on Making Light generally spawns at least two almost entirely separate conversations to keep track of. I, personally, occasionally find this difficult. It would be nice to have some way to keep track of them. Either through some sort of "thread" system, or through tagging, or something.
One minimal change might be that whenever someone responds to a post by putting "Lance@26" into their response, the ML system could automatically turn that into a link: Lance@26.
This should be pretty easy to implement, and then we'd just have to get into the habit of doing this more regularly.
What do you think?
Malthus@28*: I think ML would be an ideal incubator for community/content tool development. The community and culture are well-developed and sustainable, the signal:noise ratio in the content is very good, and the members generally seem to be receptive to the idea of introducing tooling that adds value.
* Sorry, no link, I'm too lazy! :)
Another idea is limited tagging. We'd have a very limited set of tags that we could apply thru check-boxen while posting. Like "poem" or "pun". Then some sort of option to view only those posts that have a given tag (or, e.g. for puns, not those posts).
Debbie (#27): Or is it the reader, collecting information of personal interest?
We start as (very roughly speaking) a like-minded community, but certainly not intellectual or emotional clones of one another. And moods will differ from day to day -- sometimes I can read the political threads (or the puns or the utter miscellany), and sometimes I want something else. That's when the "click on by" option that I mentioned elsewhere comes in handy.
What's important to me is a variety of ongoing discussions, by people I like and respect, so I can pick and choose. And that's what "Making Light" offers. Matchmaker programs designed to hook up happy flurosphere couples, triads etc. don't seem necessary. (And I really don't mind that nobody else here gives a damn about pro tennis or cryptic crossword puzzles! There's more to life than two minor obsessions.)
The reason I prefer limited tagging to unlimited tagging is because given, say, a post on growing roses, one person might tag it "roses", while another will tag it "flowers", and another "green-thumb", or "horticulture", or "i-for-one-welcome-our-new-perennial-overlords". Or any subset of the above and another two dozen descriptors I've left out. Makes tagging useless for looking up all posts about roses.
I don't think we can do this kind of idea justice without getting into the life cycles of conversations and communities. The needs of someone getting into a new community or area of interest will be way different from those of established members or longtime aficionados. Telling newbies to come back when they've read the FAQ (or whatever has been tagged as the most cogent exegesis of some particular subject) isn't going to give them the buzz of actually participating in a conversation.
So part of what you need to tag (if tagging is the right way) is which conversations and communities are new and growing, which are stable, which are mostly occupied by an old guard. And then you need to propagate this information in such a way that it can affect even people who don't know they need it.
I've participated in several long-lived online communities (15-20 years, long enough for people's children to be online), and have found that very few discussions are really new. Sometimes there are new points of information that didn't exist before, sometimes new casts of characters will produce a different direction, but the important
thing for the participants is having the discussion, not necessarily being able to point out the last five times or places it was perfectly hashed out. Until we can work out the "new" versus "new to me" issue, the tags will be difficult.
Somewhat off-topic.
Let me propose a corollary to Sturgeon’s Law: Everyone can agree the 90% of everything is crap, but no one will agree which 90% that is. (heresiarch #18)
This is very close to the argument I tried to make at a meeting in the Powerhouse Museum lo these many years ago including one of the influential people of the time back when the DVD region system was being set up. He was saying he & other taste-arbiters would go out into the market and pick the content *they* wanted to release in the Australia/Pacific/South American region, the European region, etc.
I was trying to say that that would wipe out the best feature of releasing so much content, old and new, on the new format, and online searching and selling; that people interested in things slightly (or well) off the "average" should now be able to have access to their interests in a far easier way, and that the total of all these different "niches" would add up to quite a substantial market, almost totally untapped at the time.
Obviously this argument didn't work on the PTB. This has not changed my belief in it.
Neil @3:
...I'd look for indicators of bad conversations, ignore them, and whatever is left over ought to be of superior quality.
Is a good conversation really just one that fails to be bad?
I've been thinking of conversational metrics in two classes; call them content and form. Many of our troll bingo filters ("you people" and the like) are content filters, conversational markers of an individual commenter who may be crossing the line, and may therefore need some moderator attention.
But our conversations tend to be resilient enough to handle a single troll. Indeed, a good piñata can liven up a discussion and spark new ideas. The markers of a conversation gone bad are probably different, more about form than content. (Though multiple trolls might count.)
An example. A few months ago, I noticed a pattern of postings that marked one way some of our threads were going wrong. (It had to do with the ratio of postings by one person to all the other participants of the conversation, with some intensifiers for comment length.) I don't have the tools at my fingertips to do the analysis programmatically, but I did use it as an informal diagnostic tool for a while.
Malthus @30 & 32:
The reason I prefer limited tagging to unlimited tagging is because given, say, a post on growing roses, one person might tag it "roses", while another will tag it "flowers", and another "green-thumb", or "horticulture", or "i-for-one-welcome-our-new-perennial-overlords". Or any subset of the above and another two dozen descriptors I've left out. Makes tagging useless for looking up all posts about roses.
This is the classic argument in favor of a controlled vocabulary, and it's a good one.
The counter-argument is that the content is analog, not digital. Meanings are gradiated, and vary according to context. And sometimes one person's terms don't make any sense to others*.
Translating things into a rigid framework too early creates data loss as well - once you over-generalize to meet the controlled vocabulary, you can't get the specificity back.
I like the idea of slowly evolving tags, kind of a semi-controlled vocabulary. So if the previous comments in the thread all talk about "roses", then "roses" is suggested, but if they're talking about "blossoms", then it's "blossoms".
A good associative search engine, with thesaurus support (I work for a company that writes and sells one for libraries, for instance) can suggest "flowers" and "blossoms" when you look for "roses".
-----
* For instance, the Library of Congress subject categorization term for books about gravestones is "sepulchral monuments". Springs right to the tongue, doesn't it?
Oh, man, there are enough issues here for several separate threads. (Making this thread a good case-study for applying curating ideas.)
Mez @34 -- those were some of the ideas rattling in the back of my head when I was thinking about who's the curator, who's the user, what are their interests, and what are their needs. Any useful system is going to have to be very flexible and robust, that's for sure.
It's been interesting observing the development of Ravelry, the knitting/crocheting megasite. From Day 1 the intention of the creators was to make it a community. They have over 55,000 members (and it's still in beta) from over 199 countries. There are pre-teens and people over 80. There are vast differences in the degree of computer-savviness, not to mention the reasons why people want to be there at all (organize their own stuff, search for yarn/patterns/ideas more easily, engage in discussions).
The discussion forums are pretty closely monitored, both by official mods and via self-policing. While there have been some complaints about trolls, I think that this is not a significant problem overall. They have a feature whereby each post to a forum can be marked educational/interesting/ funny/agree/disagree/love. I suppose you could use information like this to give some sort of overall quality rating to a discussion (I have no idea if that's the eventual goal of the site), but it would feel very clunky doing something like that at ML.
Sorry, I have no concrete ideas for solutions to any of this, but I am enjoying listening and thinking.
#32 Malthus:
This deficiency in a unlimited-tag searching system could be overcome if the search mechanism worked through complex semantic frames (see, just as an example, the Framenet project). In such a context, the use of the word "rose" (or a search on the word "rose") invokes related words/concepts such as "flower", "plant", "garden", "horticulture", "perennial" to greater or lesser degrees, depending on underlying semantic relatedness. (Of course, the use of/search on the word "rose" also invokes words and concepts related to the verb "to rise", but all this means is that a frame-based tag-interpretation system has at least a vague hope of indexing puns in a useful way.)
Bruce @15:
Automatic mechanisms for classifying or indexing poorly-understood domains strike me as problematic. What I usually ask of automation in the beginning of any major undertaking is assisting the work of an expert rather than performing all the work.
Amen. I'm a tester in my day job, and I'm well familiar with the magical thinking that goes on around automation. If you automate crap, all you get is really fast crap.
I suspect that it will be some time before automation can replace intelligent processing. There are, however, two pieces of middle ground.
1. As you suggest, software to amplify or assist human judgement. In the same way that I think you could use troll bingo markers to prioritize conversations for moderator attention, I suspect you could use a human to evaluate, steer, and teach a tagging algorithm. It might never be completed, as the conversation moves and the tags have to be made to follow. But if teaching the system is easier than labelling manually, we're still ahead of the game.
(This is not to say that this scales linearly. Assuming that a piece of software which saves a human 75% of the rote work doesn't mean they can then do the work of four people; we don't operate at our top 25% efficiency all the time. Sometimes rote work is necessarily restful.)
2. Inadequate functionality as a substitute for no functionality at all. If we wait for an optimal solution, we may wait forever. It was once described to me as "making the best the enemy of the good."
abi @ 39
on point 2: I've heard it described as
the hardest part of being an engineer is learning to say 'It's good enough: ship it.'
abi @ 39... If you automate crap, all you get is really fast crap.
I think I'll have a t-shirt made with that saying on it.
Serge @9:
Isn't the community itself a content?
Sometimes. Here, for instance, it is, partly because people (you, for instance) value it and nurture it.
Sometimes it's not. Wikipedia, for instance, uses "what contributes to the encyclopedia?" to decide what is permitted. (It comes up, for instance, in the matter of user page labels listing interests, political views, and tastes.)
Wikipedia, by the way, strikes me as being on a pretty difficult cleft stick about community vs content. Their mission is content; that's indisputable. They permit some real jerks to stay around because of the quality of their contributions.
But their policies on editor anonymity are damaging to the encyclopedia -- to content. They allow any amount of hidden conflict of interest and consequent bias. The defense of editor anonymity, due to some extreme examples of harassment and stalking, is the defense of a site that values its community. (This is not a bad thing, but it's not their stated aim.)
I don't know the answer to their dilemma. I think they're too big and too widely referenced to be immune from every kind of gaming and abuse going. They're certainly too large to be considered a single community; reputation and accountability don't scale into the numbers they have. The result is that editors and admins form cliques and cabals just to work as a community. Sadly, that damages wider trust.
I have often wondered whether they could move to some kind of federated model, but I don't know what kind.
P J Evans @ 40... 'It's good enough: ship it.'
Problems arise over the definition of good enough. Or the tester/user with the final say is technically reasonably knowledgeable, but, even though you've fixed the frigging bug, the t/u wants to know the exact circumstances that led to the buggy situation and you spend many emails telling the t/u that this wouldn't be time wisely spent and the exchange goes on and on until the t/u goes oops and realizes that he/she caused the buggy situation.
Malthus @28: One minimal change might be that whenever someone responds to a post by putting "Lance@26" into their response, the ML system could automatically turn that into a link: Lance@26.
Yes. What I've wanted for quite a while has been the other thing that could do -- which is add notes at the bottom of Lance's number-26 post that says "Replies in @28, @44" (also with links, of course).
abi @ 42... I suppose that Wikipedia, because of its size, isn't a community anymore but an organization. Maybe it hasn't admitted that to itself, or that it needs to change to actually function as an organization.
The conversation's fractal. How to find
Community may be one garden path
After another. Gardens of the mind
Require both gardeners and busy bees;
We also serve, who only pollinate
A Fibonacci series of remarks,
Roses of verse, twined on the garden gate,
An argument like fuchsia, shooting sparks,
Elaborations full as double blooms.
The vine of listening's indeterminate;
It is the green that limns these outdoor rooms.
And, should we get a blight of bile or hate,
A poison flower tipped with spiky horns
The gardener comes in nd clps th thrns
..and, of course, I forgot the comma on the penultimate line. *eyeroll*
elise, you are an ornament to our garden, a rose of splendid beauty and fragrance.
elise -- brava!!
(we need poetry tags badly; not to mention recipes, first aid techniques, book recommendations....) :-)
elise shall from now on be known as the Rose. (I see myself as a punderosa tree, providing shelter against harsh winds... or fuel for the next marshmallow roast.)
Brooks Moses @ #19: What is it, other than the fact that -- like Everest -- the possibilty of archiving these conversations is there, that makes also solved the halting problem.it worth saving them?
Well, I see your point in that nothing particularly exciting springs to mind. But if the archived conversations are available to be sifted, some hypothetical future genius might come up with some clever use that isn't obvious to either of us.
But regardless of whether anything good comes of archiving conversations, there ARE going to be downsides. The presidential candidates of 2048 are going to spend a lot of time explaining the youthfully exuberant remarks they posted to LJ and the drunken debauchery photos on MySpace.
A further musing on Brooks' point:
Even if conversations are spectacularly good, is it necessarily good to save them?
I know there have been some times when I have been weary and down, and I've been "sucked in" and spent hours browsing archived conversations on ML (ones I was part of, and ones I wasn't.) I could have been trying to participate in a new conversation, but I didn't have the energy, and it was easier to just read old ones, where I needn't take part because I couldn't.
A conversation is always a performance. I think one reason ML works is that so many people here recognize that and put a little effort into their performance. (Or sometimes a lot, as witness Elise's sonnet above.) Yet it's always easier to be a spectator than a performer, and the more that brilliant performances* are archived, the stronger the temptation to remain a spectator.
I'm not arguing against archives and curation, just musing.
[*] Bless you yet again, Mike Ford.
P.S. Joe @ 21: I think there probably are some people interested in that kind of thing who might want your archives. It seems to me there was something on Boing-Boing recently about that (IIRC, not the Pew Reports item, a different one.)
Paula @16:
Distribution and context...excellent strands to bring into this plait of ideas. A few quibbles and meditations*:
If someone's never been exposed to quality goods, or never taught what and why about "quality," or how to think critically, or is obdurate about dismissing the scientific method in favor of uncritical credo... that person is probably going to not be able to distinguish crap from quality and won't care about distinguishing crap from quality.
I think you do people a disservice here. I think that most of us have been exposed, at one point or another, to enough quality that they know it when they see it. It's simply that not everyone uses quality as their prime criterion for evaluating content; some use orthodoxy, or utility, or some other value.
We do the same in other contexts. Consider a fine handmade rug from Nepal, each of its hundreds of thousands of knots individually tied by the hands of an underpaid and exploited child. Its quality is indisputable -- such rugs last for years, even generations. It is beautiful. And I wouldn't buy it under any circumstances, because the manner of its creation is repugnant to me, and I will not support a market for such things.
Regarding community and quality, one size almost never fits much less flatters all. (Flatters might not be the word I want, the word I want denotes adornment that improves how someone appears and puts them in a positive light, highlights them positively, etc.)
It's a good term to use in balance with "fits", but a better standalone choice might be "suits".
I agree that one specific solution would never fit every community. You'd have to start with a very loose idea of what you want, and then accept that each community would move in that direction in its own ways. Finding some Grand Unified Theory of curating conversations is all very well in theory, but not really practical in real life.
Besides, it might be useful to have a little evolutionary competition and cross-fertilisation out there.
What I'm proposing above, is that context matters. "Give me a lever to push with and I will move the Earth," involves there having to be some fulcrum available for the lever, a person to push on the lever, and of course the Earth for the person to push the Earth with. The fulcrum usually goes unremarked and left out as requirement.
An excellent point.
This thread is actually, unintentionally, a worked example in the importance and unimportance of context. I originally wanted context for the quotes that Patrick posted, but the tone of the discussion got in the way. So I took them out of their context, and saw where they led me. (Actually, much of what we're noodling around here is the a wider and less commercial iteration of what Gavin Bell actually presented, but I didn't know that until the bloggers caught up with themselves.)
-----
* that would make a good name for a blog
I keep reading the parenthetical part of the title as "a mediation in the sunlight." Is that what curating a conversation is? It seems to work.
I are a doofus. By conflating two versions, I ended up trying to rhyme "path" with "bees." No "eye" and "symmetry" free pass there, I'm afraid.
OK, take two:
The conversation's fractal. How to find
Community may be one garden path
After another. Gardens of the mind
Need gardeners plus bees; you do the math.
We also serve, who only pollinate
A Fibonacci series of remarks,
Roses of verse, twined on the garden gate,
An argument like fuchsia, shooting sparks,
Elaborations full as double blooms.
The vine of listening's indeterminate;
It is the green that limns these outdoor rooms.
And, should we get a blight of bile or hate,
A poison flower tipped with spiky horns,
The gardener comes in nd clps th thrns.
heresiarch @18:
Imagine a Digg that’s designed to track individuals’ tastes. Instead of creating a single front-page list, every list is personal: as each user browses around, they tag things they like, and people they like. If a comment or a post strikes them as particularly intelligent, that alias/person (independent of site), along with the comment, is added to a list of People I Think Are Neat. If a Person I Think Is Neat posts to a blog, or some of them show up on a comment thread, the user gets notified. If the user thinks someone is really extra neat, they can go a browse that person’s list of neat things, and see if any of them appeal. It’s a system designed to privilege personal connections over group identities, individuals over sites. Each individual can create a hand-tailored community of their own, reflecting as precisely as possible their own quirks and interests.
I love this idea, both as expressed here and as touched on further on.
This requires a few things, some technical, some social. As usual, the technical elements are the easier ones to implement.
You need a consistent IDs across the web. OpenID looks like the best contender for a consistent, verified and lasting identity. There would also have to be tracking across websites and communities so you could see who was where. Here I'll just wave my hand reassuringly and mutter things about GreaseMonkey or Facebook or Web 2.0*.
The difficult thing, I think, is getting enough people willing to be followed about on the web. Some of us have sufficiently unified and distinctive identities that we can be tracked. (Patrick and Teresa, of course, but also me...Google "evilrooster" and there lies my history) But that kind of thing opens one up for stalking, identity theft, and many kinds of dreadful creepiness†. I think it will take some time before enough people get over these fears.
If good content produces good communities, and vice versa, can that teach us anything about why communities fail? Do they fail because their content sucks?
I have been watching a former community of mine wither. I even came back -- something I swore I would never do -- and Cassandra'd at them. They may yet pull through. What happened for them, anyway, is that the web changed. Other places had better stuff and lower entry thresholds. It stopped being fun. And once the people stopped coming, attrition set in. "Too many cobwebs, not enough spinners," confided one who stayed.
Let me propose a corollary to Sturgeon’s Law: Everyone can agree the 90% of everything is crap, but no one will agree which 90% that is.
Let me propose a toast to heresiarch, who has thought a very clever thing. Now if we can only keep the context of the content, generated within this community, so it is not lost...hmmm....
-----
* I hereby dub this "Strossing". Not to be confused with "Strossing out", definitions of which are invited.
† another potential blog name.
Clifton@52: It was only a little effort. I poetize like breathing, mostly. Then again, I'm asthmatic, so that might not be saying much. Mostly, though, I like Twain's remark about "exuding the stuff, as the otter exudes the precious otter of roses."
elise 56: I wondered about that rhyme! But no one else said anything, and I had a nervous suspicion that I was missing an inside joke, or something. So I kept mum (npi).
Were you taught that nonsense about "rhymes to the eye" being used in Shakespeare's day, too? Even in High School I had figured out what was really going on...'loved' and 'proved' rhymed in Elizabethan English! Rhymes to the eye, my upper back thigh.
elise @ 56 and just about everybody else, too.
First, elise, your poem is terrific, and it brings up one of the fundamental points I want to comment on. This has been a very dense thread; I want to remark on a number of basic themes that have been brought up already, and I'm expecting more.
Community value versus individual value
(Gardens of the mind
Need gardeners plus bees; you do the math.):
We're looking at a spectrum of needs and requirements here, with most of them at the two ends of individual and community. But community is fractal, though I think the requirements are not quite scale-invariant, so a spectrum is probably a good model.
An individual wants a way to record where interesting things are, find out when more interesting things are created by known interesting people, and search out more unknown interesting things and people without having to jump randomly into the web and trust to luck. An example: my default browser page is my customized Google homepage, which has some basic tools like email, calendar, weather forecast, and todo lists, and a whole slew of links and RSS feeds. When I find an interesting link or (preferably) feed, I stick it on the homepage; occasionally I remove one that's gone dormant or ceased to interest me. Right now there are just under 40 different sites with links or multi-link feeds pointed at them; I suspect I could handle fewer than twice that number with this technology before getting overwhelmed.
There are two major technological improvements I'd like to see for that general class of widget: a way to reliably filter the feeds so as to show only the items from a given site that fit my interests (so I can have more sites represented for a given amount of attention span), and some way to get the next level of abstraction: a way to automatically find links I might be interested in and present them as if they were coming from one or more aggregated feeds, say collected by type or subject.
The same sort of tools could be applied to posts, threads, comments, poster IDs, and subjects on a community like ML or a particular LJ blog. This gives the individual a way to find the most interesting stuff in the face of the proverbial internet firehose. Note that there's a short-term use for archiving here: I might discover an interesting thread days after it started, but while it is still going on; with good filtering I might be able to ignore those items in the thread I'm less interested in, and skim through until I'm uptodate enough to join the discussion. Important note: all these mechanisms should tag and weigh things from the point of view of the individual, since I don't have the same criteria, nor do I categorize things in the same way as others might.
That's bottom-up for the individual, now top-down for the community. There are global critera and categories that are important at various levels of community: a whole blog site, one thread, a branching subthread, etc. Some of them involve moderation (is this post germane to a given thread; does it meet the criteria for a civil response, etc). As with individuals, and as I said in my previous post, these things shouldn't be automatic, they should be used by the humans and the software in collaboration.
One way the individual and community interact is that a community tag could be used as the initial value (possibly modified by known mappings from this community's categories to this individual's categories) for an individual's tag for that item. And community tags in some cases might be created by looking at individuals' tags.
Design versus Growth
I agree that trying to foresee all the twists and turns a community will go through in its development is hopeless, and trying to get everything right the first time is clueless. But design is (at least in my design philosophy) more about making sure the artifacts you create are easy to extend, alter, and repurpose than about getting exactly the right implementation the first time. And you don't have to wait to ship the best, but you can incrementally ship better and better as you go along.
This is good for several reasons. First, your tools can grow and change to match the changes in the community, both scale and purpose. Second, tools can be reused in similar applications for other communities that are similar but not the same. Also, it ought to be possible to have the tools themselves track how they are used* so as to get feedback on how well they work, and what they could be extended to do.
What is a "conversation"?
One obvious example of a conversation is what we're doing here in this thread. But there are other kinds of content that fit the definition, I think. Let's take the broadest possible concept: "a conversation is a collection of atomic units of content" as a basic definition and see what fits in.
Some of you may have spent time at Shadow Unit. If you haven't seen it, it's a site created by several SFF writers as a sort of virtual TV show, with episodes (1st one was just posted last Sunday), and a vast collection of collateral material and message board threads. One neat thing is that several of the show's characters have LJs, and they blog regularly, and comment on each other's blogs. Also, some of the writers comment (the meta has been known to cause headaches throughout the web) on the blogs, and some fans have posted there as well. I call these blog threads, fictional though they may be, conversations. I expect to see more such conversations in the near-future, and things that are even more unorthodox and unexpected.
Another kind of conversation is an automatic (or semi-automatic) aggregation of items with a some common characteristics and an interest value for some individual or community. At its simplest, this might mean an aggregation of posts by bloggers that examine a common subject or theme, or that respond to each other, even though they were posted on different, unrelated sites.
I'm not going to attempt to list kinds of conversations beyond this, I just wanted to point out how diverse the notion can become.
* I know, there are privacy issues, and a lot of other nasty questions here, but it might be worth thinking about them.
Serge (#50) Rosaleo perhaps? Or some variation thereof. Referring to elise's usual preferred nom-de-net.
I think that Elise must surely be the aforesaid Precious Otter of Roses.
Serge #50: That assumes that someone is pining for you.
Fragano @ 63... Trying to derail this conversation into a punfest? Vile temptor!
Xopher @ 59: Were you taught that nonsense about "rhymes to the eye" being used in Shakespeare's day, too?
My high school (or maybe college freshman, it's been a while) English teacher told us that Blake would have pronounced "symmetry" to rhyme with "eye," and a quick listen to the Copper Family (who pronounce "lovely" as "love-lie" even when not rhyming it) would seem to back him up.
I think that Brooks' first point @19, which I'd call the Jeff Goldblum maxim*, bears some considering. Why are we doing this? Are we really going to use it? Are we going to use it enough to justify the effort? Or is it just Everest-work?
Attention, after all, is single-threaded. I may have nine‡ tabs open on my browser right now, plus two Skype chats and a TextPad document, but that's just lying to myself, pretending that opened = read = processed = internalised.
With limited attention and the fire-hose of new content always turned on and aimed at us, why would we look back at old stuff? Even the old stuff that wasn't for reading once and then just knowing, but bears rereading, takes time away from the new! shiny! content that we're all frantically typing into comment boxes right now.
I can think of a few reasons to go back, in no particular order and with no claim to completeness.
I think that the first four points are the ones that make me want to have access to older conversations. Knowledge is power, whether it be the origins of a good idea or the source of a damaging quote. I want a level playing field, where everyone can access the goods, not just the few who have the memory or resources to descend into the bit-mines looking for diamonds.
-----
* Jurassic Park: your scientists were so preoccupied with whether or not they could, they didn't stop to think if they should.
† The posting that started this thread is an example of that.
‡ Ironically, one of them is The Autumn of the Multitasker, which is about how we can't multitask very well. I think. I was halfway through it when I realised I wanted to finish this comment before reading more.
Did Abi say 'Jeff Goldblum?
Dr. Grant: Did you read Malcolm's book?
[Erik nods]
Dr. Grant: So?
Erik: I don't know. It was kinda preachy. And too much Chaos. Everything Chaos. It just seemed like the guy was high on himself.
Dr. Grant: That's two things we have in common.
Sorry, Abi, I couldn't help myself.
Heather@38:
Yes, I like that idea. When I originally posted, I had been thinking about the possible some of those toys on the Net that will, given a word, put up a graph of all related concepts in their dictionary. I think that it would be pretty interesting if one of the larger communities-with-tagging would do something with that -- but I think it would be too much work for not enough return for something the size of Making Light.
abi@36: I completely understand your point about restricted language. To a certain extent, I was talking about something incremental that could be applied here without too much work/difficulty. And on the other hand, it does express my dissatisfaction with certain styles of tagging. For example, while I find Flickr tagging to be fairly useful, I've never seen any use to the Slashdot tagging.
Tim 65: Exactly, and your teacher was better-informed than most.
I think this will turn out to be a really hard problem. And the reason I think this is because of the Amazon book recommending algorithm. When I first saw it, I was really excited. But then I realized that it mostly recommended other books by the same authors (no duh) and other books in the same genre (not very much duh). And after all these years it hasn't gotten very much better.
If my two favorite authors are Jane Austen and Terry Pratchett, I don't want recommendations for other books by them, and nor do I want recommendations of other Regency novels or fantasy.
I want books by people who make me laugh and use irony well. And how is Amazon going to figure this out? Is there someone else out there just like me? Is anyone going to be clicking the "irony" tag, other than me? I don't see how this gets solved.
Brooks Moses: And I almost never read any of these more than once.
But I do. I use things like Lj, Making Light, and any number of fora where I participte as sources of storage; for both my words, and others.
I go back and re-read old threads. Some of them have been mined by other sites (About.com uses one of my disquisitions on Interrogation as it's source).
Information is of value. One of the things I like about that is the lack of loss. There are a few people I know (or knew of) who have since died. What they wrote is still up. They can't speak anymore, but the voice wasn't stilled when they died.
I have a large amount of writing which is on paper, and unless those APAs, letter, newspaper articles; columns, cards, etc. are collected somehow, they are dead to the world. Some are archived, so as long as 1: the archivist keeps them, and 2: the paper survives (there are two-copies, archived of every word which went into my college paper, one to the EIC for that term, one in the morgue),they survive.
But absent some outside referent, who will know to look for people who kept copies of Myriad when I was active? Or my HS school paper?
But if someone wants to look into the attitudes toward interrogation in the early oughts of the 21st century, they can find a lot of what I wrote. Maybe that will point them to other writings (to see what shaped my present thinking).
Re content: We keep being told we are in a "visual medium now" I keep wanting to ask people who tell me this means words aren't all that needful, because this is the "MTV Age" (though this is much less common than it used to be) to remove the speakers from their televisions, and see how comprehensible the programming is.
elise: I used to be more poetic (though sonnets have never been my meter), but of late I've not been reading as much, and so the muscles have witherèd. I am grateful you (and so many others) don't so suffer.
re 45: Wikipedia acts like a community, but in the sense that, say, California or China is a community. Which is to say, there is some commonality and assumed openness to interaction based upon participation in an overall shared purpose, but the thing is simply too big to function as a community as everyone imagines by default. But it isn't really an organization either, for the same reason that California isn't an organization.
abi @ 66
I think there are a number of good reasons to want to archive access to a conversation. One I mentioned in my previous post is based on the notion that conversations on the web need not be synchronous, so I may want to catch up on a conversation that started before I found out about it.
Another is that, as I also mentioned, the notion of conversation is potentially much broader than the definition we've been giving to it. I might very well be interested in a conversation that occurred some time ago because of the people (real or otherwise) who participated in it. And a conversation could in fact be a synthetic aggregate of content that was never before considered as one single unit.
You're right that attention is a scarce resource, and needs to be conserved and divided optimally among as many interesting things as possible. But I wonder if we can't have some technological aid there. I've spent a little time looking at that in the past, and I think it would be possible to build an assistant that could learn priorities of particular links, feeds, whatever, and order the display of them to match. It could base priorities on interest, time since last viewing, relation to other links recently visited, etc.
Bruce Cohen @ 66.... I think it would be possible to build an assistant that could learn priorities of particular links, feeds, whatever, and order the display of them to match
Hopefully the assistant won't be built by Ron Goulart.
Serge #64: Not at all, since you are intent on policing the punctilio of discussion here, I'm certainly not going to lead us into a pun-fest.
I think, however, that we need to consider that conversation is as much (and perhaps more) about communing as it is about communicating.
Bruce @74:
Short comment, because it's gone 1am for me. I'll do more tomorrow.
And a conversation could in fact be a synthetic aggregate of content that was never before considered as one single unit.
That sideswipes something I've noticed on Making Light itself.
We think we're running multiple parallel conversations; what I write here is related to what you've written above. But there's horizontal crossover as well. We deal with each other in multiple threads at the same time. I've seen quarrels spill from one thread to another; we all have. More subtly, I've watched anger build silently in one thread and explode in another, to some bafflement.
That too is a conversation we don't consider - the entirety of Making Light as a single entity, as well as a collection of smaller ones.
Serge #77: You have me confused with my old friend S.A. Tanas, who was expelled from Paradise as an invertebrate punster.
Though this is an obvious response to Brooks's remarks in #20, I hope it's not a trivial one.
So, then, my question: What is it, other than the fact that -- like Everest -- the possibilty of archiving these conversations is there, that makes it worth saving them?
[...]
Beyond that, do conversations preserve their worth beyond the interaction of those participating?
As many of you know, I have been researching Robert Heinlein's history.
The collaboration between Heinlein as author and John W. Campbell, Jr. as editor, beginning in 1939, is pivotal to a change in science fiction that affected all subsequent history.
Heinlein lived in California (at first), JWC lived in New Jersey, so much of their relationship was postal.
When books of Campbell's letters came out, they had practically none of his correspondence with Heinlein. Grumbles from the Grave, a selection of Heinlein's letters, had a few exchanges with JWC, but left me unsatisfied.
Heinlein donated his papers to the University of California at Santa Cruz. Last year, his estate made scanned copies available for a modest fee. So I bought all the Heinlein-Campbell letters, four boxes' worth. I was not disappointed.
These guys wrote letters, sometimes long ones, several times a week, for several years. They start stiff and formal, but after a few stories have been submitted, they get more chatty. They talk about editorial changes, of course. They talk about word rates. They recommend books to one another. They talk about their hobbies. They tal
Comments on Curating conversations (a meditation in the sunlight):