1 hour ago, Eiuol said:
Why? I mean, I _do _expect that content will be controlled, precisely because messages must be altered so that they can be transmitted. A message has to be converted into binary and then an electric signal, then re-translated back. This isn’t a mere expression of the message, but the very content of it. Details must be abstracted away, or otherwise altered. The telecommunications company must decide for themselves how best to alter the message so that you will be a satisfied customer, and even decide whether content should be modified on the linguistic level. This way, the system works as intended.
You’re talking about the encoding of the message, like deciding whether to use UTF-8 or UTF-16. That is a separate issue.
Some lower-level mechanisms might encode bits into analog pulses using QAM constellations and things like that. There’s also the use of error-correcting codes and compression. For text, lossless compression is probably best, but for audio or video it’s also possible to employ various “lossy” compression mechanisms, like converting speech into low bit rates for cellular phone systems, which is done using a model of the kinds of sounds that commonly occur in speech. (It’s also possible to use a model of the kinds of sounds the human ear can distinguish, which leads to MP3 and other similar codecs.)
However, you’ve already noted that “the system works as intended” only as long as the content of the message is preserved, and that is precisely what I am arguing in favor of. There was a scandal a while ago about certain copiers which used a lossy compression algorithm and which would replace numbers on architectural drawings with other numbers – thereby producing invalid copies of originally valid drawings. This was a scandal for precisely that reason.
If Stephen King wrote a novel and somebody used a text predictor such as GPT-3 to lossily compress it, and then expand it, and the result was more or less like the original novel but with some of the words and sentences changed, this would upset Stephen King, because he has the right to decide what goes under his byline, and it might upset his readers, too, who have the right to expect that something “by Stephen King” was actually written by him. Unless somebody checks, there’s no way to know whether the machine made a substitution that actually changed the meaning. (Even though the algorithm would get even better compression if it could do things like eliminating that janitor in chapter 32 who doesn’t really do anything important to the plot, things like that.)
An “AI” algorithm making judgments about what to allow through and what to block would resemble “lossy” compression. Both kinds of systems are capable of error, too; in compression this gives rise to “compression artifacts.”
1 hour ago, Eiuol said:
The problem is, the motivating issue for you isn’t false representation, or that companies modify content in such a way that nobody knows what another person actually said. No one is doing this. You are talking about deleting posts or suspending accounts, things like that. In other words, your argument is a total red herring.
There’s this thing called “shadow banning” where the company purports to have delivered your message but really hasn’t. So you might think that someone received your message when this is not the case.
That can mess things up if you tell your friend that you’ll pay for the pizza, but your friend never receives this message and so wonders why you are so rude… but it can have other implications, too. If there’s a controversial court ruling, one might ask, why hasn’t {favorite political commentator} commented about this yet? Does {commentator} agree with the ruling or not? Why the stony silence? Is {commentator} simply too shocked to say anything?
It’s possible for a social media company to “quote you out of context” by banning messages.
This is easiest if they can delete some posts but allow others through, such as by turning shadow banning on and off. Consider if a publisher was able to omit about half the words from a book; they could make the book say completely different things with the other half of the words, although it would be a shorter book.
It is also possible to misrepresent someone’s speech by preventing them from speaking further. (Like if a publisher deleted the last few chapters of a book).
For example, if you write a post in the heat of the moment, which may have errors, and someone declares it objectionable on some basis or other, and you scratch your head about this, but decide that maybe you should post a follow-up and clarify with further statements that you are not talking about that objectionable thing – but you now can’t, because you’ve been banned. So if you made a mistake you cannot correct it, and if they made a mistake you cannot point it out. In either case, your actual point of view is not properly represented, and if you were about to say that someone making mistakes while they write “in the heat of the moment” is blaming others for their own mistakes, that’s only partially true, since a person might write differently if he knows that a post is going to be his last. Context is important in a matter like that.
I suppose I have slipped from the idea of a self-contained “message” to something more like a “conversation,” which is an ordered collection of messages, and has context. But the main idea – preserving the integrity of the message (or conversation) itself – still holds.
There’s also an issue with the separation of state and economics. This separation does not have force of law in any country that I’m aware of, but it should.
Whenever two parties enter into a contract, there is a question of what court has jurisdiction over it – and there should always be some court that has jurisdiction, so that in the event of a dispute, the matter can be resolved.
I think the US Supreme Court erred by allowing companies to write binding arbitration into contracts instead of using the courts, because binding arbitration is doing the job of a court system without making the same guarantees about the rights of the accused and so forth that a real court would be obliged to make. It violates the separation of state and economics. (Binding arbitration is used to solve a problem that should have been solved by other means.)
An even worse violation of the separation of state and economics is when a court upholds a contract clause like “XYZ corporation shall be the sole determiner of whether the user is in breach of this contract or not.” That would mean that XYZ corporation is setting itself up to replace the court’s judgment and to simply instruct the court to impose damages. The court should strike down clauses like that.
Further, you have the right to a public trial, and that means that the evidence against you is a matter of public record. That means the opposite of what the censorship people want: if you write a “bad” post, the post should be preserved, because it is evidence. (Of course if it violates someone’s privacy, or has other genuinely dangerous properties, such as telling terrorists how to disable the fire suppression system in a particular building, a judge could order it redacted accordingly).
Generally, however, court actions are public record, and this enables people to read rulings and decide for themselves if the rulings were fair.