The Wikipedia killer
Posted: Fri Mar 30, 2018 2:54 am
As I've mentioned elsewhere, if there is to be a killer app that destroys Wikipedia, it isn't likely to be an encyclopedia at all, but some kind of technology that accepts natural language queries, contextualise and summarises all relevant reliable sources on the fly, and outputs the answer in a format most relevant/convenient for the customer. Be that an article, table, list or one word. ("No" obviously)
This will of course only be something the big tech giants can achieve, and since Wikipedia has killed the commercial market for simply providing knowledge, it's likely to be packaged as a freebie with one of their pay for services/products.
Once customers are assured the result will be comprehensive and accurate, and most likely even neutral once the tech is refined further, then the demand for the often incomplete, inaccurate and biased Wikipedian derived knowledge disappears overnight. As such, where studies of Wikipedia's quality have been sparse, limited and even contradictory, stress tests of this tech will be widespread.
It seems likely the tech will always struggle to be as polished as the best Wikipedia article could be if given enough volunteer time. But in realty consumers are already wise to the fact only a tiny amount of what is on Wikipedia is what you would generously consider polished. Indeed, since the tech would be able to follow a Manual of Style and do other basic shit like not repeating itself way better than Wikipedia can in most articles, it likely won't perform so badly even in the polish stakes.
And of course there are two huge advantages of this tech that will absolutely crush Wikipedia - it will always be giving you the most up to date knowledge, and it will never give you information that someone just made up one day, for a laugh.
The threat posed by citogenisis, Wikipedia's greatest gift to humanity, while potentially disastrous to the accuracy of results here, would be managed by some combination of the on the fly process, plus blacklisting and good old fashioned detective work, perhaps even conducted as a crowd sourced effort. How quaint.
For those who think this is a far out pipe dream, consider the pace of technology, and what is already being worked on as we speak.....
https://www.theregister.co.uk/2018/02/1 ... wikipedia/
....sadly (and surprisingly), the people at El Reg mistake this as a potential future tool for Wikipedians to use to help them write articles, rather than the future basis for their potential replacement.
HTD.
This will of course only be something the big tech giants can achieve, and since Wikipedia has killed the commercial market for simply providing knowledge, it's likely to be packaged as a freebie with one of their pay for services/products.
Once customers are assured the result will be comprehensive and accurate, and most likely even neutral once the tech is refined further, then the demand for the often incomplete, inaccurate and biased Wikipedian derived knowledge disappears overnight. As such, where studies of Wikipedia's quality have been sparse, limited and even contradictory, stress tests of this tech will be widespread.
It seems likely the tech will always struggle to be as polished as the best Wikipedia article could be if given enough volunteer time. But in realty consumers are already wise to the fact only a tiny amount of what is on Wikipedia is what you would generously consider polished. Indeed, since the tech would be able to follow a Manual of Style and do other basic shit like not repeating itself way better than Wikipedia can in most articles, it likely won't perform so badly even in the polish stakes.
And of course there are two huge advantages of this tech that will absolutely crush Wikipedia - it will always be giving you the most up to date knowledge, and it will never give you information that someone just made up one day, for a laugh.
The threat posed by citogenisis, Wikipedia's greatest gift to humanity, while potentially disastrous to the accuracy of results here, would be managed by some combination of the on the fly process, plus blacklisting and good old fashioned detective work, perhaps even conducted as a crowd sourced effort. How quaint.
For those who think this is a far out pipe dream, consider the pace of technology, and what is already being worked on as we speak.....
https://www.theregister.co.uk/2018/02/1 ... wikipedia/
....sadly (and surprisingly), the people at El Reg mistake this as a potential future tool for Wikipedians to use to help them write articles, rather than the future basis for their potential replacement.
HTD.