I missed that era. Could you explain a bit more about webs of trust in google reader?
tag jokes
Wouldn’t be tolerated, generally. Arborists would simply not have weed in their curation nets.
I’m sort of open to using subjective processes to assign tags their names. In this case, communities would tend to name their own tags, but outsiders, who find the name opaque or ambiguous, wouldn’t have to put up with it. One thing that would enable is translating tags into different languages.
automatic generation of tags
I think how this will happen is, if a curation web splits, or forms an isolated component, some mostly automated ministry process will detect that, issue a recommendation to define a new tag or split a tag into more specific subgenres. Votes are taken, and the community decides whether they want to do that based on the results.
Weights
Meaning, degree of applicability? No. Tags would generally be predicates with clear meanings. Where different gradations of a quality need to be registered, multiple tags should be made along a continuum. I kind of want to write a lot about why taking the average of big crowds of error-ridden scores isn’t usually (unless you’re estimating the weight of a cow at a county fair) a great way of getting at an underlying truth about any real objective score they’re supposed to reflect.
how tags are protected
When someone tags incorrectly, the people who endorsed them are supposed to unendorse them. If they fail to do that, their endorsers may unendorse them in turn.
There would emerge a few components in the scholarly consensus web that are dens of horseshit, but the presences in the (we hope) largest one will proactively disconnect themselves from those. The people who have real uses for a scholarly consensus clique would have little use for liars on the fringes, so once identified they’d mostly end up being ignored.
Wouldn’t be tolerated, generally. Arborists would simply not have weed in their curation nets.
If you’re intending this as a top down decree I could see it, but if you’re making a prediction about how people would use an open-ended system I think you’re wrong.
To clarify, stoners would misuse the tree tag, people looking for actual tree content wouldn’t see any of that because they would not be using the stoner web to sort results.
There were no explicit webs of trust in Google Reader. (I think that might be better, actually.) But ‘following’ someone else would give you access to everything they ‘shared’ (on Google Reader), including things shared by other people they followed (after which you could follow that third person), and so on recursively.
tag jokes
Wouldn’t be tolerated, generally. Arborists would simply not have weed in their curation nets.
I’m sort of open to using subjective processes to assign tags their names. In this case, communities would tend to name their own tags, but outsiders, who find the name opaque or ambiguous, wouldn’t have to put up with it. One thing that would enable is translating tags into different languages.
I can’t tell if you’re assuming/expecting ‘arborists’ will entirely volunteer their services or whether they’d be officially and explicitly recognized by your proposed system.
I think ‘tag translation’ would be very interesting – not just between languages (‘with an army’) but between jargon and plain language too.
automatic generation of tags
I think how this will happen is, if a curation web splits, or forms an isolated component, some mostly automated ministry process will detect that, issue a recommendation to define a new tag or split a tag into more specific subgenres. Votes are taken, and the community decides whether they want to do that based on the results.
Weights
Meaning, degree of applicability? No. Tags would generally be predicates with clear meanings. Where different gradations of a quality need to be registered, multiple tags should be made along a continuum. I kind of want to write a lot about why taking the average of big crowds of error-ridden scores isn’t usually (unless you’re estimating the weight of a cow at a county fair) a great way of getting at an underlying truth about any real objective score they’re supposed to reflect.
how tags are protected
When someone tags incorrectly, the people who endorsed them are supposed to unendorse them. If they fail to do that, their endorsers may unendorse them in turn.
There would emerge a few components in the scholarly consensus web that are dens of horseshit, but the presences in the (we hope) largest one will proactively disconnect themselves from those. The people who have real uses for a scholarly consensus clique would have little use for liars on the fringes, so once identified they’d mostly end up being ignored.
I’m skeptical that this is feasible with tags intended to be “predicates with clear meanings”. I just don’t think that’s generally how people use tags/words/concepts. I’d think it’d be both easier for you, or the system implementor, and the system’s users, if tags were entirely ‘open ended’. I think the killer feature of what you’ve outlined is the explicit ‘web of trust/taste/curation/consensus’. I don’t think trying to pin tags to fixed static meanings can work even over the ‘medium term’.
I missed that era. Could you explain a bit more about webs of trust in google reader?
Wouldn’t be tolerated, generally. Arborists would simply not have weed in their curation nets.
I’m sort of open to using subjective processes to assign tags their names. In this case, communities would tend to name their own tags, but outsiders, who find the name opaque or ambiguous, wouldn’t have to put up with it. One thing that would enable is translating tags into different languages.
I think how this will happen is, if a curation web splits, or forms an isolated component, some mostly automated ministry process will detect that, issue a recommendation to define a new tag or split a tag into more specific subgenres. Votes are taken, and the community decides whether they want to do that based on the results.
Meaning, degree of applicability? No. Tags would generally be predicates with clear meanings. Where different gradations of a quality need to be registered, multiple tags should be made along a continuum. I kind of want to write a lot about why taking the average of big crowds of error-ridden scores isn’t usually (unless you’re estimating the weight of a cow at a county fair) a great way of getting at an underlying truth about any real objective score they’re supposed to reflect.
When someone tags incorrectly, the people who endorsed them are supposed to unendorse them. If they fail to do that, their endorsers may unendorse them in turn.
There would emerge a few components in the scholarly consensus web that are dens of horseshit, but the presences in the (we hope) largest one will proactively disconnect themselves from those. The people who have real uses for a scholarly consensus clique would have little use for liars on the fringes, so once identified they’d mostly end up being ignored.
If you’re intending this as a top down decree I could see it, but if you’re making a prediction about how people would use an open-ended system I think you’re wrong.
To clarify, stoners would misuse the tree tag, people looking for actual tree content wouldn’t see any of that because they would not be using the stoner web to sort results.
There were no explicit webs of trust in Google Reader. (I think that might be better, actually.) But ‘following’ someone else would give you access to everything they ‘shared’ (on Google Reader), including things shared by other people they followed (after which you could follow that third person), and so on recursively.
I can’t tell if you’re assuming/expecting ‘arborists’ will entirely volunteer their services or whether they’d be officially and explicitly recognized by your proposed system.
I think ‘tag translation’ would be very interesting – not just between languages (‘with an army’) but between jargon and plain language too.
I’m skeptical that this is feasible with tags intended to be “predicates with clear meanings”. I just don’t think that’s generally how people use tags/words/concepts. I’d think it’d be both easier for you, or the system implementor, and the system’s users, if tags were entirely ‘open ended’. I think the killer feature of what you’ve outlined is the explicit ‘web of trust/taste/curation/consensus’. I don’t think trying to pin tags to fixed static meanings can work even over the ‘medium term’.