With the average opsec of chat sites, would you bet on "effectively"? Ironically Parler itself just had a leak.
Even if a company currently has good opsec, it can be acquired and go downhill (or turn evil).
In the current climate, this comment could (and would) be construed as implicit support of Parler, with all the consequences. The rich left loves "consequences", except for themselves.
But suppose you set up a specific semi-public organization to act as the identity provider. (Like IANA? Maybe that's not a good example. :-)) Following something like a provably secure communications flow[1], the chat site gets a unique number representing the user. That's all they know about the user.
Alternatively, and with the approval of the user, the site gets the user's name or whatever the user wishes their online reputation to be associated with and can display that, for their comments to be taken seriously.
Ok, so let's take the example of Parler. Parler doesn't moderate its content, and actively resists attempts outside attempts to do so. However, anything illegal is still associated with individuals and the court system can force Parler to complete the link (that's the "effectively anonymous" approach, here), resulting in that user facing legal consequences. (This is opposed to the current system, where Parler could be technically designed to be unable to complete that link (Yeah, Parler are idiots.) and any kind of moderation relies on Parler's goodwill.)
AWS can continue to host Parler under the cover of a statement like "The POPO are keeping them in check." Or, AWS can kick them off based on measurable facts like "they got more than 10 or 15 subpoenas a day for three months".
As for your last paragraph...
Personally, I think Parler was a sewer filled with the morally and intellectually challenged who lack any sense of responsibility as citizens of the United States of America. (My understanding is that its users were limited to the US, btw.) I understand and agree with its vendor's decisions not to do business with them any more, in the same way I understand and agree with their decisions not to do business with ISIS recruitment sites.
The rich left loves "consequences", except for themselves. The rich right loves "consequences", except for themselves. The poor love "consequences", except for themselves. Libertarians love "consequences", except for themselves. You love "consequences", except for yourself. I love "consequences", except for myself.[2] Everyone loves "consequences", except for themselves. Because humans suck.
[1] Yeah, I'm a formal logic guy.
[2] Except I don't really, schadenfreude aside. I seem find in myself a streak of moralism and an excessive sensitivity to fairness. It sucks. Even my schadenfreude doesn't extend to the Darwin awards.
No, wait, hear me out.
No moderation isn't acceptable. Really. (If someone wants to make a counter-assertion, I merely point to this site.)
Non-transparent, corporate moderation doesn't seem palatable to anyone. There are just too many pitfalls.
Independent moderation falls apart when you consider that different fora have different moderation requirements. (A forum for Cinderblock the obese cat is going to be very different from any kind of political or technical site.)
The best case seems to me to be case-by-case, transparent moderation with precedents, similar to common law. And ultimately, I expect transparent moderation to, in the extreme, go to the court system, so that's not necessarily a bad starting point.
Unfortunately, that falls apart when large numbers of participants are (including, say, moderators) are not speaking in good faith. I see two possible ideas to address that: First, to tie accounts to physical identities (but not necessarily disallowing the account to be effectively anonymous), to cut the number of bots, multiple accounts, etc. Second, to attach an account to a user's reputation (which does break anonymity). The results I see are to cut down on the volume of crap while ensuring users have skin in the game (with a side order of making legal action against stalking, harassment, and threats).
But what about those situations that require anonymity? Most of those already have legal and social protections: psychological, religious, and legal counseling, for example. I would support anonymity for those fora, which places responsibility for moderation on the fora, of course, as well as meaning that the moderation cannot be transparent.
One area does require special handling: whistleblowing. It does require anonymity, and does not have any current legal or social protections. That needs to be fixed.
But anyway, as a general rule, the default for social media should not support anonymity. The only way to free providers like Facebook, Reddit, this site, or AWS from responsibility for what is posted there is to place that responsibility on the actual posters---having no responsibility doesn't work, and giving that power to the discretion of the owner of the provider isn't acceptable.
There are some objections that I think I can answer, but this comment is getting too long for me, so I'll wait until anyone cares.
And no, the irony of Parler's requirement of photo-ids and (allegedly) SSNs isn't lost on me.