Oct 05, 2010
Erin McKean, Wordnik
The language is the Dictionary
If you took the language, and you got rid of the dictionary, what would be left would be awesome!
Language is a shared delusion that we all have - each word that we use means - and only means whatever we agree. In order for a word to really mean something, we have to agree with other people - words do not mean whatever you want them to mean.
We can get words from wherever we like. We get them from other languages. We mix and match parts. And as long as we agree, it's a word. And, you can't unword a word once people start to use that word as a word. Nobody can force you to use a word - that's the beauty of a shared delusion.
Wittgenstein - the meaning of a word is its use in a language.
So where do we go to find the meaning of a word? A dictionary? Yes - that's where most people go to. But - Dean Trench - talked about the meaning of words - paper presented to the Philological Society on deficiencies in the dictionary. "Let's find a bunch of uses of words, and get as much information of the word as we can." And that's how dictionaries have been created ever since.
If you think of it - a dictionary definition is a very lossy representation format. The definitions are taken from actual uses. And the contributors constitute (a limited) social network. A limited number of people to get information from.
Interestingly - the more a person wants to know a word, the less likely it's in the dictionary. People want definitions of the rare, the new, the usual - exactly the words you don't expect to find in tne dictionary (survey: most people who bought dictionaries bought them more than five years ago).
Back to Wittgenstein - why not just get rid of the 'lossy filter' and just show me the sentences where it occurs. If a word is important enough to somebody, and you look long enough, someone will have defined it somewhere. 'geeksta rap' for example. Or 'graphere' - a close cousin of carbon nanotubes. Invented in 2004. If you look hard enough, somebody will give an explicit definition of the word somewhere.
On Wordnik we try to show you as many different sentences as we can (up to 20), ranked according to how well the sentence makes clear the meaning of the word. Interestingly - journalists tend to be better dictionary editors than most dictionary editors - there's something at stake, so they do a good job.
(Because words are defined in the sentence they're in we have this idiosyncratic ideas of meaning - even the common words where we think we agree on the meanings of we don't. (My view too)). Using 'the meaning' in the same way as 'the dictionary'. Discussion of metaphors - metaphors are well all the fun is. In dictionary editors - we have 'lumpers' and 'splitters' - trying to join or split definitions of words.)
What about made up words - like 'madeupical'? They are like placeholders - something's going to be here eventually. "Of course it's real! After all, I made it up, and I'm real! Case closed. Besides, no one ever puts madeupical words on Wordie. Ever"
(What about words that are better to use, or shouldn't be used. -- think of words like that as social constructs. Like attitudes toward drunk driving. We've changed the social construct around that sort of behaviour. Or, consider hats. Or pantyhose. Or ties. Or bowties. Or XKCD's fake Wikipedia article about a fake word. malamanteau. Newness counts.
But where's the awesome? It's in looking at the traffic of where people use words and looking at what the patterns are. What sio the GPS of a word? The reason words are better than planets is that words change more. If you have all these words and you know what they're doing - what would you do with ti? Maps are inherently worth having. What would you do with it? erin@wordnik.com
(Where is the shared meaning? It's in the way we agree to use the word - the conventions - like we don't say "yes" when someone says "Can you pass the salt.") It's kind of like a distributed authority.
(SDÂ - this was a remarkable fascinating witty talk, much more so than is apparent from this summary).
Riley Crane, MIT Media Lab
The World Is becoming More Responsive (but it's not there yet)
Talk is about how we're not quite there yet, but how we can get there. In particular, talking about how crowdsourcing can make things more responsive.
An early example of crowdsourcing - a trail in the woods. We are seeing the digital equivalent of this in things like Google Instant. It's a form of crowdsourcing.
These forms of crowdsourcing do not solve problems that require coordination or collaboration. They are taking some aggregate statistical picture of likes or dislikes, and using that to make suggestions to us. They are enabled by the great evolution of the link - from ARPANet to the WWW to the Social Web. Evolution of the link.
Can we design systems that leverage the new hyper-connected web to solve really hard problems. ARPAnet did the '10 red balloons' challenge. We learned about it four days before it was announced. What is the role of the internet and social networking in solving these real-time communication challenge. How quickly can you mobilize society?
Duncan J. Watts, et al. "We conclude that although global social networks are in principle searchable, actual success depends sensitively on individual incentives."
Basically, they build a pyramid scheme (the network is a classic tree, with levels, etc). Nothing really happened to start. Then it was picked up on Slashdot. Then we got some results. This is a social cascade. (we hear a tone version of the cascade). The network - in 36 hours, 5000 participants.
So - how did we win the competition? We basically built a spontaneous human sensor network. Easy, right? If we had been foresightful, we would have had systems to manage the data coming in. We found that there's a lot of sabotage in social networks. (people thought Media Lab was doing military-funded research; others were from other teams).
Information was filtered in various ways - there was eg. information in the noise - eg. valid reports would have many different accounts, not exactly the same account. Or we could detect photos - not photoshopped versions. Or just a simple filter to take information only from people whose IP filter says they were near where they reported the balloon to be. And at the end of the day, we used a whiteboard to aggregate all the information. Often what we need are toolkits, not tools - you can't always scope things out ahead of time.
In the end, it took us 8:52 minutes to find all 10 balloons. Half the people who found balloons were people who signed up first and found a balloon later. Incentives are important, but perhaps social capital is more important.
(Comment - the key seems to have been being picked up by Slashdot - could that be engineered - it was engineered - we planned that - also being picked up by a CNN reporter - we sent links to anyone who talked about the red balloon challenge).
(SD - what this shows is that we're still in a mass media environment & whomevr has access to mass media still wins.)
Could this be extended? To find say people who can operate backhoes? Foind missing children? Etc. Or an example from the Haitian earthquake that needed coordinates for the locations of people who were trapped, using Ushahidi.
The story of the balloons, etc. shows: our communication paradigms are broken. If you think about these examples - they cobbled together a communication network, they cobbled together a bunch of tools to create a spontaneous communication network. We can distinguished between strong ties, weak ties, but also spontaneous ties (or temporary ties - like the bus drivers, or all of you in this room - we're spontaneously connected, on a temporarily connected).
Twitter and Facebook have started to solve some of these problems with our communication paradigms. Facebook tackled the problem of keeping track of weak ties - people didn't want to go through all that effort. Facebook lowered the cost of communicating with our weak ties. Twitter solved the problem of sending a message to everybody you know - but we allowed anyone who is interested to follow. They solved the problem (?) of spontaneous communication. The next breakthrough will be solving the problem of digital communication with temporary ties.
John Perry Barlow, EFF and TTI Vanguard
Reflections
What happened today was something that so often happens, where a subject that seemed familiar, was filled with things that hadn't been seen before - things really right on the frontier, where there are still the fundamental human problems, how to deal with authority, what is trust, what is true, what rings true, and how we deal with reality in the most efficient way, especially some that comes in huge flat chunks of data.
Riley said our communications paradigms are broken. We knew - in 1990 - that we would be flooded with information, and we would go into data shock. But just as we navigate the phenomenal worls, we've evolved the ability to make a pretty reasonable sort. Now, each of us is part of the sort. We need to be a part of the engine of doing that.
Gladwell piece - the revolution won't be tweeted. It seemed grumpy, but some of the things he said were difficult to argue with. I was involved in helping people, set up proxy servers, etc., but i couldn't sort out what was important, and in the end the revolution did not happen. A girl getting her cellphone returned wasn't really impressive.
But what we have is the ability to form very brief, but useful, ties, and to turn very weak ties, into strong ties. (Digression into an explanation of ball and spring maps - interesting that everybody talked about something that was map-like - not maps of geographies but spaces that we don't understand).
(Discussion about the day)
(SD - my own impression is that we got a lot of surface phenomea today, but not so much in detail.)
What I would have said, given a chance...
What we've seen today is mostly surface phenomena - that's why there's so much use of maps - and ver little ability to see the deeper phenomena.
Eg. what constitutes 'agreement'? How do the networks actually produce knowledge / insight /awareness etc.
This is complicated by the almost unremitting desire of people in the room to attempt to interpret these according to old-world values and explanations - things like 'power', 'trust', 'incentive', 'propagation' or 'broadcasting' or 'messaging'.
Also complicating any analysis is the mix of actual examples of peer production, and what may most generously be called 'assisted' peer production. A rock band uses a marketing firm to generate buzz. MIT Media Lab uses Slashdot and CNN to get a boost. Cluster analysts use humans to interpret the meanings of terms and to screen data. Etc. These are cases where old media creates results, but where these results are interpreted to be a consequence of peer phenomena.
What contribution does the network make? There's no evidence that people even see this as a question yet.
(I had someone say, "it's more efficient to think of a task problem as being addressed by the running of a script in the mind, as opposed to the very inefficient neural net mechanisms I describe. But this simply offloads the problem. What is running the script? )
(And someone else said, "where is the impact of all of these technologies? When will the GDP go up?" And I thought, that's exactly the wrong question - we want the GDP to go down, because all these things we used to do that were so expensive, are now nearly free.)
day two - and my own talk - tomorrow.
The language is the Dictionary
If you took the language, and you got rid of the dictionary, what would be left would be awesome!
Language is a shared delusion that we all have - each word that we use means - and only means whatever we agree. In order for a word to really mean something, we have to agree with other people - words do not mean whatever you want them to mean.
We can get words from wherever we like. We get them from other languages. We mix and match parts. And as long as we agree, it's a word. And, you can't unword a word once people start to use that word as a word. Nobody can force you to use a word - that's the beauty of a shared delusion.
Wittgenstein - the meaning of a word is its use in a language.
So where do we go to find the meaning of a word? A dictionary? Yes - that's where most people go to. But - Dean Trench - talked about the meaning of words - paper presented to the Philological Society on deficiencies in the dictionary. "Let's find a bunch of uses of words, and get as much information of the word as we can." And that's how dictionaries have been created ever since.
If you think of it - a dictionary definition is a very lossy representation format. The definitions are taken from actual uses. And the contributors constitute (a limited) social network. A limited number of people to get information from.
Interestingly - the more a person wants to know a word, the less likely it's in the dictionary. People want definitions of the rare, the new, the usual - exactly the words you don't expect to find in tne dictionary (survey: most people who bought dictionaries bought them more than five years ago).
Back to Wittgenstein - why not just get rid of the 'lossy filter' and just show me the sentences where it occurs. If a word is important enough to somebody, and you look long enough, someone will have defined it somewhere. 'geeksta rap' for example. Or 'graphere' - a close cousin of carbon nanotubes. Invented in 2004. If you look hard enough, somebody will give an explicit definition of the word somewhere.
On Wordnik we try to show you as many different sentences as we can (up to 20), ranked according to how well the sentence makes clear the meaning of the word. Interestingly - journalists tend to be better dictionary editors than most dictionary editors - there's something at stake, so they do a good job.
(Because words are defined in the sentence they're in we have this idiosyncratic ideas of meaning - even the common words where we think we agree on the meanings of we don't. (My view too)). Using 'the meaning' in the same way as 'the dictionary'. Discussion of metaphors - metaphors are well all the fun is. In dictionary editors - we have 'lumpers' and 'splitters' - trying to join or split definitions of words.)
What about made up words - like 'madeupical'? They are like placeholders - something's going to be here eventually. "Of course it's real! After all, I made it up, and I'm real! Case closed. Besides, no one ever puts madeupical words on Wordie. Ever"
(What about words that are better to use, or shouldn't be used. -- think of words like that as social constructs. Like attitudes toward drunk driving. We've changed the social construct around that sort of behaviour. Or, consider hats. Or pantyhose. Or ties. Or bowties. Or XKCD's fake Wikipedia article about a fake word. malamanteau. Newness counts.
But where's the awesome? It's in looking at the traffic of where people use words and looking at what the patterns are. What sio the GPS of a word? The reason words are better than planets is that words change more. If you have all these words and you know what they're doing - what would you do with ti? Maps are inherently worth having. What would you do with it? erin@wordnik.com
(Where is the shared meaning? It's in the way we agree to use the word - the conventions - like we don't say "yes" when someone says "Can you pass the salt.") It's kind of like a distributed authority.
(SDÂ - this was a remarkable fascinating witty talk, much more so than is apparent from this summary).
Riley Crane, MIT Media Lab
The World Is becoming More Responsive (but it's not there yet)
Talk is about how we're not quite there yet, but how we can get there. In particular, talking about how crowdsourcing can make things more responsive.
An early example of crowdsourcing - a trail in the woods. We are seeing the digital equivalent of this in things like Google Instant. It's a form of crowdsourcing.
These forms of crowdsourcing do not solve problems that require coordination or collaboration. They are taking some aggregate statistical picture of likes or dislikes, and using that to make suggestions to us. They are enabled by the great evolution of the link - from ARPANet to the WWW to the Social Web. Evolution of the link.
Can we design systems that leverage the new hyper-connected web to solve really hard problems. ARPAnet did the '10 red balloons' challenge. We learned about it four days before it was announced. What is the role of the internet and social networking in solving these real-time communication challenge. How quickly can you mobilize society?
Duncan J. Watts, et al. "We conclude that although global social networks are in principle searchable, actual success depends sensitively on individual incentives."
Basically, they build a pyramid scheme (the network is a classic tree, with levels, etc). Nothing really happened to start. Then it was picked up on Slashdot. Then we got some results. This is a social cascade. (we hear a tone version of the cascade). The network - in 36 hours, 5000 participants.
So - how did we win the competition? We basically built a spontaneous human sensor network. Easy, right? If we had been foresightful, we would have had systems to manage the data coming in. We found that there's a lot of sabotage in social networks. (people thought Media Lab was doing military-funded research; others were from other teams).
Information was filtered in various ways - there was eg. information in the noise - eg. valid reports would have many different accounts, not exactly the same account. Or we could detect photos - not photoshopped versions. Or just a simple filter to take information only from people whose IP filter says they were near where they reported the balloon to be. And at the end of the day, we used a whiteboard to aggregate all the information. Often what we need are toolkits, not tools - you can't always scope things out ahead of time.
In the end, it took us 8:52 minutes to find all 10 balloons. Half the people who found balloons were people who signed up first and found a balloon later. Incentives are important, but perhaps social capital is more important.
(Comment - the key seems to have been being picked up by Slashdot - could that be engineered - it was engineered - we planned that - also being picked up by a CNN reporter - we sent links to anyone who talked about the red balloon challenge).
(SD - what this shows is that we're still in a mass media environment & whomevr has access to mass media still wins.)
Could this be extended? To find say people who can operate backhoes? Foind missing children? Etc. Or an example from the Haitian earthquake that needed coordinates for the locations of people who were trapped, using Ushahidi.
The story of the balloons, etc. shows: our communication paradigms are broken. If you think about these examples - they cobbled together a communication network, they cobbled together a bunch of tools to create a spontaneous communication network. We can distinguished between strong ties, weak ties, but also spontaneous ties (or temporary ties - like the bus drivers, or all of you in this room - we're spontaneously connected, on a temporarily connected).
Twitter and Facebook have started to solve some of these problems with our communication paradigms. Facebook tackled the problem of keeping track of weak ties - people didn't want to go through all that effort. Facebook lowered the cost of communicating with our weak ties. Twitter solved the problem of sending a message to everybody you know - but we allowed anyone who is interested to follow. They solved the problem (?) of spontaneous communication. The next breakthrough will be solving the problem of digital communication with temporary ties.
John Perry Barlow, EFF and TTI Vanguard
Reflections
What happened today was something that so often happens, where a subject that seemed familiar, was filled with things that hadn't been seen before - things really right on the frontier, where there are still the fundamental human problems, how to deal with authority, what is trust, what is true, what rings true, and how we deal with reality in the most efficient way, especially some that comes in huge flat chunks of data.
Riley said our communications paradigms are broken. We knew - in 1990 - that we would be flooded with information, and we would go into data shock. But just as we navigate the phenomenal worls, we've evolved the ability to make a pretty reasonable sort. Now, each of us is part of the sort. We need to be a part of the engine of doing that.
Gladwell piece - the revolution won't be tweeted. It seemed grumpy, but some of the things he said were difficult to argue with. I was involved in helping people, set up proxy servers, etc., but i couldn't sort out what was important, and in the end the revolution did not happen. A girl getting her cellphone returned wasn't really impressive.
But what we have is the ability to form very brief, but useful, ties, and to turn very weak ties, into strong ties. (Digression into an explanation of ball and spring maps - interesting that everybody talked about something that was map-like - not maps of geographies but spaces that we don't understand).
(Discussion about the day)
(SD - my own impression is that we got a lot of surface phenomea today, but not so much in detail.)
What I would have said, given a chance...
What we've seen today is mostly surface phenomena - that's why there's so much use of maps - and ver little ability to see the deeper phenomena.
Eg. what constitutes 'agreement'? How do the networks actually produce knowledge / insight /awareness etc.
This is complicated by the almost unremitting desire of people in the room to attempt to interpret these according to old-world values and explanations - things like 'power', 'trust', 'incentive', 'propagation' or 'broadcasting' or 'messaging'.
Also complicating any analysis is the mix of actual examples of peer production, and what may most generously be called 'assisted' peer production. A rock band uses a marketing firm to generate buzz. MIT Media Lab uses Slashdot and CNN to get a boost. Cluster analysts use humans to interpret the meanings of terms and to screen data. Etc. These are cases where old media creates results, but where these results are interpreted to be a consequence of peer phenomena.
What contribution does the network make? There's no evidence that people even see this as a question yet.
(I had someone say, "it's more efficient to think of a task problem as being addressed by the running of a script in the mind, as opposed to the very inefficient neural net mechanisms I describe. But this simply offloads the problem. What is running the script? )
(And someone else said, "where is the impact of all of these technologies? When will the GDP go up?" And I thought, that's exactly the wrong question - we want the GDP to go down, because all these things we used to do that were so expensive, are now nearly free.)
day two - and my own talk - tomorrow.
