4 ms·

The thing that worries me even more is that more and more diverse types of content will be susceptible to algorithmic amplification in more and more sophisticat

The thing that worries me even more is that more and more diverse types of content will be susceptible to algorithmic amplification in more and more sophisticated ways. Right now it's still relatively easy to tell what is a bot and what isn't because it's hard to generate credible text and video (and many other kinds of) content, but that's rapidly changing. It's a game of cat and mouse clicks. This pervades every single sphere of life. Soon, it will become more and more difficult to distinguish the truthfulness of digital representations. In Baudrillard's words, I suspect that big data is becoming hyperreal. It is not that data is misrepresenting reality, but that it is actively influencing it, changing it to fit its representations. Furthermore (and somewhat orthogonal), distinguishing fake content is getting harder because you don't even necessarily need bots (technological capital). If I started a restaurant and I wanted to acquire the maximal amount of attention and I had no moral qualms, I could hire 10,000 MTurks (or MTurk-equivalents) to write reviews for me on Yelp, and it will drive a positive feedback loop of legitimate good reviews. (This is exactly what's happening with Amazon rankings and reviews.) Amplification isn't just algorithmic, it can be produced by the application of many different kinds of capital.

zby 8y ago
We need to work on our assumptions - why human content is supposed to be good and bot content is supposed to be misleading? We assume that people have good intentions - but, as you notice, this easily breaks if evil organizations can pay people to create the content. In the past page rank was (partially) solving this content problem - but now http pages are not the important nodes any more. I would welcome new developments in page rank like algos - but more localized, because smaller networks are easier to police and evolve. Unfortunately it looks that https://en.wikipedia.org/wiki/Advogato https://en.wikipedia.org/wiki/Advogato is dead now.
- gambler 8y ago
  >I would welcome new developments in page rank like algos PageRank-like algorithms are not the solution, they are the problem. They reward popularity with more popularity, enabling viral content, encouraging stuff like link spam and outrage click-bait. Worst of all, they normalized the idea that what you see in search results (and later - timelines, feeds) shouldn't be based on your query and author's page content, but rather on what everyone else wants to see.
  - ryandrake 8y ago
    there are lots of current day Internet-scale problems where, in order to solve them, someone is going to have to figure out how to overcome people’s natural tendency to mistake popularity for quality. This falsehood compromises all kinds of rankings, from search results to Yelp and Amazon reviews to tweets. Even here on HN, “number of upvotes” (popularity) unsatisfactorily attempts to declare something about a comment’s quality.
    - harimau777 8y ago
      I think part of the problem is that in many circumstances, such as Yelp and Amazon reviews, popularity is the best proxy for quality that is currently available. Perhaps if other ways for people to quickly distinguish quality were developed then people would start making better decisions.
- rauhl 8y ago
  > We need to work on our assumptions - why human content is supposed to be good and bot content is supposed to be misleading? For some reason I’m reminded of https://xkcd.com/810/ https://xkcd.com/810/ Really, we want to reward good information. I think that there may be a philosophical issue here: not everyone agrees on what ‘good’ is (maybe noöne agrees!): I may derive much joy from Y but you find it derivative and lame; you may love X but I hate it; he may enjoy passing around Russian agitprop memes. I don’t know what the answer is, but I think neither ‘the majority of people in this bubble in San Francisco like this’ nor ‘ the majority of people on earth like this’ is a terribly useful, not to mention correct, criterion.

zby 8y ago

We need to work on our assumptions - why human content is supposed to be good and bot content is supposed to be misleading? We assume that people have good intentions - but, as you notice, this easily breaks if evil organizations can pay people to create the content. In the past page rank was (partially) solving this content problem - but now http pages are not the important nodes any more. I would welcome new developments in page rank like algos - but more localized, because smaller networks are easier to police and evolve. Unfortunately it looks that https://en.wikipedia.org/wiki/Advogato https://en.wikipedia.org/wiki/Advogato is dead now.

gambler 8y ago
>I would welcome new developments in page rank like algos PageRank-like algorithms are not the solution, they are the problem. They reward popularity with more popularity, enabling viral content, encouraging stuff like link spam and outrage click-bait. Worst of all, they normalized the idea that what you see in search results (and later - timelines, feeds) shouldn't be based on your query and author's page content, but rather on what everyone else wants to see.
- ryandrake 8y ago
  there are lots of current day Internet-scale problems where, in order to solve them, someone is going to have to figure out how to overcome people’s natural tendency to mistake popularity for quality. This falsehood compromises all kinds of rankings, from search results to Yelp and Amazon reviews to tweets. Even here on HN, “number of upvotes” (popularity) unsatisfactorily attempts to declare something about a comment’s quality.
  - harimau777 8y ago
    I think part of the problem is that in many circumstances, such as Yelp and Amazon reviews, popularity is the best proxy for quality that is currently available. Perhaps if other ways for people to quickly distinguish quality were developed then people would start making better decisions.
rauhl 8y ago
> We need to work on our assumptions - why human content is supposed to be good and bot content is supposed to be misleading? For some reason I’m reminded of https://xkcd.com/810/ https://xkcd.com/810/ Really, we want to reward good information. I think that there may be a philosophical issue here: not everyone agrees on what ‘good’ is (maybe noöne agrees!): I may derive much joy from Y but you find it derivative and lame; you may love X but I hate it; he may enjoy passing around Russian agitprop memes. I don’t know what the answer is, but I think neither ‘the majority of people in this bubble in San Francisco like this’ nor ‘ the majority of people on earth like this’ is a terribly useful, not to mention correct, criterion.