(C) Daily Kos
This story was originally published by Daily Kos and is unaltered.
 . . . . . . . . . .



Section 230 from a geek's perspective [1]

['This Content Is Not Subject To Review Daily Kos Staff Prior To Publication.']

Date: 2023-11-02  

This started out as a response to this article about Facebook's algorithm driving hatred, but it got long enough that I decided it was time to actually try writing a diary for once. So here goes.

I’ve been around the net for longer than Section 230 has been a thing (maybe not a lot longer, but longer). I spent a while being one of the plumbers who keep the tubes in working order, the big ones that let you talk to people somewhere outside your own office or neighborhood. And I see a fundamental problem not with the way Section 230 was written and intended, but with how it has been applied. In no small part this is because at the time it was written one of the core drivers of the problem was not really a big enough thing to be readily visible yet, and when it became so, the courts failed to properly understand a crucial distinction.

Very simply: not all algorithms are created equal.

An algorithm that acts solely in response to human inputs and returns its best understanding of what the human is asking for (your normal search engine, at least on a good day) is very fundamentally different in terms of its goals and operation than a selection algorithm such as is found on current social media. This is true at the technical level, but in a much more relevant way it is visible at the business level. The key question that needs to be asked is “what drives someone to use the service and thus provide potential revenue (either subscription or ad-based)?”

For a search engine, the answer is roughly “return the most useful answer to the question being asked”. Yes, there are some warts on this, but by and large unless the engine is being overtly manipulated in ways that are fairly easy to document the biases it has in the output will be overwhelmingly determined by the inputs. Manipulating the selection of inputs is also relatively easy to detect. But in general the value of a search engine goes up the more information it has access to, so there is a fairly non-trivial business driver to feed it as much as possible. Google may no longer have “try not to be evil” as a motto, but most of the peril from them is about how much data they collect to target advertising and how it can be used, rather than the fundamental algorithm that drives the search engine. At the end of the day, they don’t really care much about how you feel about what they showed you, as long as it was what you asked for, and other than the clearly labelled paid results, they don’t have any real driving reason to bias the results.

This is, roughly speaking, part of what what Section 230 was intended to protect, and not terribly dissimilar from what was contemporary at the time — an aggregation of speech by a variety of speakers, and in so far as the search engine is expressing any “opinion” at all, it is strictly a function of what the user requested. The opinion is only “this seems like it matches what you asked for”.

Originally social media sites had relatively simplistic algorithms that weren’t all that different. In fact, you can still see some remnants of that era in various places on many of them. But their business model has always been to engage the audience, to keep them looking at the site for as long as possible (rather than asking as many questions as possible). And from that stems a multitude of problems. But one of the biggest is that what they show you is no longer a disinterested opinion of “here is what seems like the best match” or even “here is what you asked for”. Instead, it is an overtly and profoundly interested opinion, who’s goal is convince you to stay engaged, whatever that takes. And it is well documented that one of the best ways to keep people engaged is to tell them things that outrage them.

The fact that courts treat both of these the same way is the crux of the problem, because they are nothing alike. If we have to treat both of them the same, then they should both be treated as published opinions by the company offering the site. Because while annoying, “the opinion expressed is simply that these results are the best fit to the question asked” is relatively easy to defend as something that should not incur liability, and “the opinion is one that encourages engagement to the exclusion of any other considerations” is rather less easy to defend.

[END]
---
[1] Url: https://www.dailykos.com/stories/2023/11/2/2203277/-Section-230-from-a-geek-s-perspective?pm_campaign=front_page&pm_source=latest_community&pm_medium=web

Published and (C) by Daily Kos
Content appears here under this condition or license: Site content may be used for any purpose without permission unless otherwise specified.

via Magical.Fish Gopher News Feeds:
gopher://magical.fish/1/feeds/news/dailykos/