Conversation

Notices

1. Embed this notice
   J. Nathan Matias 🦣 (natematias@social.coop)'s status on Monday, 15-Apr-2024 23:58:43 JST J. Nathan Matias 🦣

   • Josh Braun

   @josh is dropping some awesome insight on the causes of technical accidents/disasters at this National Academies workshop.

   Josh has summarized the issues in this *really great* video on costs and incentives for advertising, news, and disinformation.

   https://vimeo.com/930828854

   • Embed this notice
     Josh Braun (josh@sciences.social)'s status on Monday, 15-Apr-2024 23:58:34 JST Josh Braun

     in reply to

     • Michelle Amazeen
     • Janet Vertesi

     @cyberlyra @natematias @commscholar I agree and I rake them over the coals in my first article on adtech. I think both things can be true at the same time, though, and agree I didn't emphasize this in the video: They _created_ a system with a horrendous risk profile because they focused on profit to the exclusion of harms. There's still plenty of culpability here.

   • Embed this notice
     Janet Vertesi (cyberlyra@hachyderm.io)'s status on Monday, 15-Apr-2024 23:58:35 JST Janet Vertesi

     in reply to

     • Josh Braun
     • Michelle Amazeen

     @josh @natematias @commscholar Yep, I just didn't want to let them off the hook. When the tech is black boxed it's easy to say, oh this risk is a property of the system and we can't avoid it, as opposed to we accepted this risk and built it in because we thought the profit outweighed the harms.

     pettter repeated this.
   • Embed this notice
     Josh Braun (josh@sciences.social)'s status on Monday, 15-Apr-2024 23:58:36 JST Josh Braun

     in reply to

     • Michelle Amazeen
     • Janet Vertesi

     @cyberlyra @natematias @commscholar This is a spot on critique! They're treating harms to the public good as acceptable risks rather than critical issues. I think on another level we're in agreement here, though, in the sense that I'm arguing that we've learned from Perrow and others that there are ways to take these problems seriously if companies were so inclined (or were forced by regulators to do so).

   • Embed this notice
     Janet Vertesi (cyberlyra@hachyderm.io)'s status on Monday, 15-Apr-2024 23:58:39 JST Janet Vertesi

     in reply to

     • Josh Braun
     • Michelle Amazeen

     @josh @natematias @commscholar this is an interesting take, thanks! I’m no stranger to NAT, but I’d also point out that when all the services started consolidation of data streams for the purposes of selling ads (in the wake of the financial crisis), many of us told them it was a bad idea for all the reasons it has turned out to be, in fact, a bad idea— bubbles, bias, privacy, misinfo. That they did it anyway speaks less to “accidents” or systemic properties than higher regard for profit motive.

   • Embed this notice
     Josh Braun (josh@sciences.social)'s status on Monday, 15-Apr-2024 23:58:40 JST Josh Braun

     in reply to

     • Michelle Amazeen

     @natematias Also, since you've been so kind as to mention this presentation, I should note that I also published an essay on this topic:

     "Normal Accidents in the Digital Age: How Programmatic Advertising Became a Disaster"

     Here's the author preprint, but I also have a PDF of the finished essay that I can give out on request to anyone who's interested:

     https://umass-my.sharepoint.com/🅱/g/personal/jabraun_umass_edu/EapFDwihCOZAqyoCIUX8JigBgN98pYMH_VvI7ItMF2NL7w?e=ErEXH8&download=1

     @commscholar has an excellent essay in the same volume (edited by Matthew P. McAllister and Emily West).

   • Embed this notice
     J. Nathan Matias 🦣 (natematias@social.coop)'s status on Monday, 15-Apr-2024 23:58:41 JST J. Nathan Matias 🦣

     in reply to

     Josh cites Charles Perrow's work on "Normal Accidents: Living with High-Risk Technologies."

     Perrow argues that multiple, unexpected failures are a basic feature of complex, integrated tech systems.

     These "normal accidents" they cascade through the system to create harm. This happens when:
     - the system is complex
     - the system is tightly coupled
     - it has catastrophic potential

     And platforms/algorithms/misinfo have all of that.

     https://en.wikipedia.org/wiki/Normal_Accidents