If you want to regain the pseudo-privacy of physical spaces online, put your community’s conversations behind an authentication barrier and disable full text search of conversations.

Search engine indexing is what turns a pseudo-private space into a humiliating-public one.

It’s okay to let search engines index your forum’s existence, the sub forums it contains, and their descriptions. But do not let them index participants or conversations - either by subject, by participants, or by content. And do not offer full text search of post content to authenticated members. It’s okay to index keyword tags, but that’s it.

If you do this, you will regain the semi-anonymity that made the early Internet possible to enjoy. If you don’t, you will continue to suffer the trolls and abuse that full-text search enabled in the mid-90s (see also DejaNews, X-No-Archive: Yes, and Google’s purchase of DejaNews).

EDIT: If you truly feel that full-text search is so valuable that it must not be withheld, you have to do a lot of things to defend against abuse attackers - for example: charge money for search credits, deduct credits when they choose to reveal the text of results, warn users that their searches will be monitored for abuse, require users to be in good standing with paid membership and posting activity for at least 90 days, etc. Otherwise trolls will just use stolen cards to perform full content searches to identify users to harass and then report their findings back to a central forum. They may still do that after all the above criteria, but they’ll have to work excruciatingly hard at it. Yeah, they could manually scrape the site, but you can defend against that too (“you’ve participated on 12 days, so you’re allowed to view 12 days of old content” is a good simple test).

But, even if you disable search, disable history, there's the fundamental fact that _anyone_ can record everything they see, easily and silently. You can't just have a private authenticated space, you need to be able to personally trust every single person you let in that space.

At that point, the features around archiving or search are a bit moot.

- We shouldn't bother checking for characteristics of credit card fraud at transaction time, because a determined attacker might get a fraudulent card through.

- We shouldn't bother checking IDs at bars, because a determined attacker might get a fake ID through.

- We shouldn't bother trying to prevent email spam, because a determined attacker might get a spam message through.

- We shouldn't bother making laws against recording people without their consent, because a determined attacker might do so anyways.

Please construct a more plausible argument than "it's ultimately hopeless". I'm willing to consider alternatives, but I'm not willing to consider fatalism.

I don't argue that "it's ultimately hopeless". I think, with effort, it's quite possible to create a pseudo-private safe space online.

I just think that the bulk of your benefit comes from vetting the intentions and judgement of your participants, and not from technological means like removing search.

If you're able to trust your participants, technological means can be like a simple fence/lock, keeping honest people honest, but don't otherwise add a ton of extra benefit.

If you're not able to trust your participants, removing search will help _some_, but it might not be enough.

If it was just "determined attacker can do a search, and use the results privately" it wouldn't be a big deal. But they can then spread the result to everyone else in the world in a way that almost negates the barriers.

The recording option is closer, but search can be done retroactively, which makes a huge difference.

Under US law anyways, it is absolutely not okay to republish a written conversation without permission under copyright law to do so, unless you fall under the "fair use" exemptions. It is safe to say that most forms of attack would not pass a "fair use" test in the courts. So if more people were to sue over copyright violations and demand the unmasking of who posted and/or those who operate, that would certainly go a long ways towards resolving that imbalance.

Sadly, it's a very expensive process, and it's very cheap to violate copyright.

If you permit full text search of content, you permit abusers to discover and harass your members, because they can just sign up to find out who posted something once full text search discovers it.

You're implying they're somehow inversely related. A site can be well represented even when it's users contents are not on display

I feel like you're right about some communities, and wrong about others, and it's interesting to distinguish the two, because I don't think this is a distinction anyone usually bothers to make.

There are some communities where the same community divides its activity across multiple channels. Your average "same small group of people, different channels" Slack or Discord server is this way. IRC communities also usually end up this way after they grow to sufficient size, forking off channels of #foo-offtopic, #foo-announce, #foo-help, etc. phpBB forums are/were well-known for their structure of forums with subforums (where most forum admins would set up even more subforums than anyone needed, just because they could) but where there were certainly always separate "news" and "chat" and "on-topic" forums.

But other "communities" (more like societies, I suppose?) like Reddit, or Usenet, or Twitter, do basically none of this constraint-based splitting. You'll get topic-based splitting, but this doesn't change the tone of the conversation at all. It's less like being in a separate place with its own rules, and more like just having your conversation tagged with a topic so that people can find conversations like that.

I find that the only time this type of community/society seems to work, is when it generates entirely coincidental non-connected member subgraphs, i.e. when its members aren't just a random sampling of the larger community/society's membership, but rather mostly their own cultural enclave that happens to use the community/society's social network as a gathering place. Then they can have (probably mostly implicit) rules that are different from the free-for-all of the larger society's.

There are also [sub-]communities with specific explicit rules, like Wikipedia, or /r/AskHistorians/. I feel like these aren't really relevant to the question, because the explicit rules often cause a selection effect in the membership who bothers posting, such that it's not much different to just picking those particular people and saying that only they can post. So you can't really use them as an example of how to solve the problem of general Internet discourse being shitty.

Can you speak more concretely what you mean here?

Are you talking about technical "constraints" into protocols such as "http" or "TCPIP"? Or constraints into DNS? Or constraints on HTML markup language?

What would an "internet technical architecture designed to prevent negativity" actually look like? Is there an example repo on github or a computer science research paper showing the algorithms that would satisfy this ideal?

Or to take another example of constraints -- if someone in a bar spreads a false rumor, that misinformation can quickly spread to all the patrons in the bar, but it's reach beyond that bar will be slow. By that time more factual information might also be circulating, and the damage of the false information blunted. Online, by contrast, misinformation spreads so much faster than factual information that it is often nearly impossible to counter.

I don't know what the solution exactly is here, but I feel that public spaces need to have more speedbumps. In the same way that people are jerks when they drive and the answer is often "less driving, and slower," I suspect that the answer to bad online social spaces is "less online, and slower," but I'm not sure what that looks like.

Kind of like how HN limits us to a handful of posts per day before the “you’re posting too much” roadblock, causing you to really think about what you have to say and whether you want to burn one of your budget on it. Usenet could have used something like that.

I think you're being a bit optimistic about physical spaces here. Rumor mills are as old as the hills. News spreads more slowly IRL, but that goes for truth as well as falsehood. You claim that misinformation spreads faster online than truth and imply this is different from IRL info, but I don't see why this should be the case. The same underlying reasons that favourite one message over another operate in both domains.

You're assuming that people will react to those signals by stopping (or not starting in the first place). Many of the most disruptive people look for those signals as their goal, because they want to be disruptive.

Think about people who will interpret any variation of "ugh" signals as "oooh, I've found the buttons to poke to get fun noises". If you want to solve this problem, that's a large part of the threat model.

The economic models we've seen so far online are different -- the product is "be social here" and I think that's problematic. Few people want to pay just to hang out, online or off. But those selling this space have to make money somehow -- so if you're Facebook, you make your money by advertising to the people hanging out at your place, meaning to make more money you need to get more people to come be advertised to, or convince advertisers they're getting more value per ad (and so you start intrusively data mining your advertising targets).

It's more like having thousands of parallel NYCs, each one focusing on just a specific subset of the overall culture. To your point, they're still public spaces though where anyone can come in and yell about their thing, that is undeniable (and it happened).

Except people are a product of context, so putting them into a new context creates a new breed, the new yorker, of which no human being aspires to.