4 ms·

I'm not sure I see the point in saying that a particular comment should or should not be hosted at a particular place. What is the actual win? Without canonica

by taeric 5y ago

I'm not sure I see the point in saying that a particular comment should or should not be hosted at a particular place. What is the actual win? Without canonical names/homes for data, much of this dream feels out of reach. And with it, we are just back to citing data in a textual way. That is, we have it already. And the canonical step is hampered by the cost of storing information. Publications largely push this to archival services. But nobody is going to want to archive every random page. Especially for the comments, which are often mutable. I similarly don't see the advantage of requiring so much semantic marking of textual data. Indeed, I'd go so far as to say that grammar is largely superfluous. Exceedingly so if you are into the esoteric rules of the game. Certainly structure is advantageous, but most structure is inferred. And anything that slows the publishing of data is likely as not to just cause it to get lost.

echelon 5y ago
> But nobody is going to want to archive every random page. Ephemeral is good enough for most and a paradigm shift here might do a lot of good to address an often unspoken problem. Having services keep hold of everything for all time is simply unsustainable. While volumes of data might have future value in creating ML models, the vast majority of it will never be accessed again. Things that deserve preservation, that have interested stakeholder(s), will be retained.
- taeric 5y ago
  Agreed. But you would still want traceability. Such that knowing where it was is still important, even if it is not there anymore.
__MatrixMan__ 5y ago
The problem with coupling content too tightly to a particular place is that it places too heavy a burden on that place to support all the different types of users that might want to interact with the content. For instance, I like to hang out in math twitter. I wish I was taking classes from these professors, but instead I just think about their tweets and ask myself: "that's a neat claim, could I prove it?" Twitter isn't the place for getting feedback on proofs, and I don't really want to hijack the thread with an external link, so I just do it on paper and throw it away. But what would be cooler is if I could create a link between my proof and the conversation that inspired it. Maybe some other aspiring student will notice it and start a dialog with me where we can learn from each other. Sure, I could post a link to overleaf or some other place that supports LaTeX collaboration but now anybody who wants to collaborate must go create an identity on that platform, and now we have the same problem all over again if somebody else, for instance, wants to play around with it via proof assistant software. You get this proliferation of third parties who have to manage a permissions mess re: who can read/append/edit what and how to render it to users with different preferences--to say nothing of alterior motives on the part of the platform.

by taeric 5y ago

echelon 5y ago

> But nobody is going to want to archive every random page. Ephemeral is good enough for most and a paradigm shift here might do a lot of good to address an often unspoken problem. Having services keep hold of everything for all time is simply unsustainable. While volumes of data might have future value in creating ML models, the vast majority of it will never be accessed again. Things that deserve preservation, that have interested stakeholder(s), will be retained.

taeric 5y ago
Agreed. But you would still want traceability. Such that knowing where it was is still important, even if it is not there anymore.

__MatrixMan__ 5y ago

The problem with coupling content too tightly to a particular place is that it places too heavy a burden on that place to support all the different types of users that might want to interact with the content. For instance, I like to hang out in math twitter. I wish I was taking classes from these professors, but instead I just think about their tweets and ask myself: "that's a neat claim, could I prove it?" Twitter isn't the place for getting feedback on proofs, and I don't really want to hijack the thread with an external link, so I just do it on paper and throw it away. But what would be cooler is if I could create a link between my proof and the conversation that inspired it. Maybe some other aspiring student will notice it and start a dialog with me where we can learn from each other. Sure, I could post a link to overleaf or some other place that supports LaTeX collaboration but now anybody who wants to collaborate must go create an identity on that platform, and now we have the same problem all over again if somebody else, for instance, wants to play around with it via proof assistant software. You get this proliferation of third parties who have to manage a permissions mess re: who can read/append/edit what and how to render it to users with different preferences--to say nothing of alterior motives on the part of the platform.