Your skeet can't rely on that "fact". Your skeet was an attempt to prove that "fact." You have causation backwards.
Okay! This is a really helpful comment for me in clarifying where I’ve been communicating badly. There’s a key claim that I’ve been implicitly operating with, that’s allowed me to try and go in both directions, and “get causation backwards”.
So, to make that key claim explicit:
S230 provides procedural benefits to service providers if and only if there exists some set of facts where liability exists under 1A, but no such liability exists under S230.
Here, I’ll use this claim to explain the skeet in question, and address the questions you raised. I assume you’ll be skeptical of this claim, so I’ll provide an argument to support the claim below.
(As an aside, because this is an if and only if claim, then you can see how I was using this in both directions: using the existence of procedural benefits to show that S230 actually eliminates some liability, and in a different context using the fact that S230 eliminates some liability to demonstrate how procedural benefits arise from it)
For now, assume the claim above is true (again, I’ll defend it below if you’re skeptical). I’ll use that claim to explain how the skeet you’ve asked about
If your claim is true, that a recommendation cannot be defamatory, then the platforms wouldn’t need 230 immunity for their algorithmic recommendations
https://bsky.app/profile/ncallaway.bsky.social/post/3kvacpr3tor2m
So, to run through the argument that I was trying to make: if a content recommendation cannot be defamatory under 1A, that means there does not exist a set of facts that could be alleged about a content recommendation that would survive a motion to dismiss.
Therefore, (applying the claim), there is not a procedural benefit that S230 provides. If S230 doesn’t provide a procedural benefit, then the platform doesn’t need it in the status quo, because it doesn’t exist in the status quo.
and yet they rely on it
However, we all generally agree that service providers have relied on the procedural benefits of S230 for the algorithmic recommendations. Therefore (again, applying the claim), there must be a hypothetical liability that exists under 1A that does not exist under S230.
You claimed that if the immunity didn't include things that otherwise would have been defamatory, then there was no point to the immunity. But that isn't the only possible point of immunity. Immunity shut down lawsuits that were clearly invalid, but would be costly to defend. ICSs would win lawsuits over user speech, but they still had to defend them. The shield made it so they could kick the invalid lawsuits straight away.
So, you and I both completely agree that Section 230 provides these kinds of procedural benefits that you’re talking about. My point is that the mechanism S230 uses to provide those procedural protections arise from eliminating a slice of liability. As I defend below, If Section 230 doesn’t eliminate any potential liability, then it simply wouldn’t provide the procedural protections that we both agree exist.
That is most definitely an absolute need of the liability shield that does not require the shield protecting anything that wasn't already protected.
Such a liability shield can exist. But Section 230 does not work that way.
As I mention below, most (all?) anti-SLAPP laws work the way you’re describing. They provide procedural protections without modifying the ultimate liability for any specific set of facts, but they do so by modifying civil procedure.
Okay, first let’s label some claims that we’ll use. You don't have to accept these claims now, I just want to define and label them, since we'll use them repeatedly throughout.
- No Direct Procedural Benefits (
NDP) - S230 does not modify civil procedure in any way to directly create a procedural benefit for service providers. - Procedural Benefits (
P) - S230 provides procedural benefits to service providers - Liability (
L) - There exist some set of facts where liability exists under 1A, but no such liability exists under S230
For the sake of being able to constrain the argument, I’m going to focus on the main procedural benefit that comes up with Section 230—the ability to dismiss a lawsuit earlier than when a claim is brought under 1A. Specifically, 230 will allow many claims to be dismissed at a Motion to Dismiss, when under 1A many of those claims might survive a motion to dismiss, go through discovery and then be dismissed at Summary Judgment.
There are other procedural benefits in S230, and I can defend others in a similar way, but I think it’s more useful to focus on a single concrete example since this response is long enough already. If you want to discuss other procedural benefits aside from dismissing cases early, I'm happy to do so.
The general structure of the argument will be to show that NDP is true, that NDP → (P ⟺ L), which then shows that P ⟺ L.
To show NDP → (P ⟺ L), I’ll show that NDP → (P → L), and that L → P.
This point is mostly here, because it’s a necessary element for the rest of the argument, but the demonstration is just a simple reference to the text of the law: https://www.law.cornell.edu/uscode/text/47/230. Section 230 isn’t particularly long, and you can read it and notice that it doesn’t modify civil procedure in any way. It doesn’t create new motions, doesn’t shift evidentiary burdens, doesn’t direct Courts what specifically to consider as they evaluate cases.
For comparison, anti-SLAPP provisions are a kind of legislative shield that are usually (always?) implemented by modifying civil procedure. Here’s California’s anti-SLAPP law: https://leginfo.legislature.ca.gov/faces/codes_displaySection.xhtml?sectionNum=425.16.&lawCode=CCP. Notice that it does all of the things I mentioned above.
(a) First, let’s re-arrange this to (NDP & P) → L.
(b) For this argument we can assume that S230 does not modify civil procedure in any way to directly create a procedural benefit for service providers, and that S230 does provide procedural benefits (again, for this argument, we’ll examine specifically the ability to dismiss cases early at a Motion to Dismiss).
(c) Let’s establish with some important civil procedure facts:
- A Motion To Dismiss can be filed immediately after the complaint is filed, and must assume all the alleged facts in the complaint are true.
- If the alleged facts in the complaint would not lead to liability, then the motion to dismiss is granted, and the case is dismissed (potentially with the ability to refile an amended complaint)
- If the alleged facts in the complaint would lead to liability, then the motion to dismiss is denied, and the case proceeds to discovery
- If the case proceeds, discovery occurs (which is the often a very expensive phase of a trial)
- A Motion for Summary Judgment can be filed after discovery is complete
- If the evidence produced by discovery clearly shows some of the critical alleged facts in the complaint cannot be supported by the evidence, the motion for summary judgment will be granted
- If the evidence produced by discovery plausibly supports the alleged facts, the motion of summary judgment is denied and the case proceeds to trial
- If the plaintiff is able to prove the alleged facts at trial, the jury will find the defendant liable
(d) So, because of P, we know that there exists a hypothetical complaint that under 1A would survive the Motion to Dismiss phase, but would be dismissed by a Motion to Dismiss under 230.
(e) The hypothetical complaint is a set of alleged facts that could either be true or false.
(f) If the set of facts in the hypothetical complaint end up being accurate, then the complaint would survive summary judgment, go to trial, and a jury could eventually find the defendant liable.
(g) So, for this hypothetical complaint, the set of facts it alleges would produce liability under 1A, but would not produce liability under S230.
(h) Therefore, S230 allows service providers to dismiss some complaints in a motion to dismiss that would otherwise advance through discovery to summary judgment.
(a) Because of L, we know we can craft a hypothetical complaint that alleges a set of facts that, if true, would lead to liability under 1A that would not lead to liability under S230.
(b) Therefore, we can craft a hypothetical complaint that would survive a motion to dismiss under 1A, but would be dismissed under S230
(c) Therefore, S230 allows service providers to dismiss complaints in a motion to dismiss that would otherwise advance through discovery to summary judgment.
So, above we’ve shown:
NDPNDP → (P → L)L → P
Therefore, we can conclude that:
NDP → (L → P)- applying 3NDP → (P ⟺ L)- applying 2, 3P ⟺ L- applying 1, 5
There’s obviously a lot here, and I’m not claiming I’ve comprehensively proved every sub-point, so if there are areas you want me to address in more detail on, just let me know.