News:

SMF 2.1.4 has been released! Take it for a spin! Read more.

Main Menu

What do people do about PMs?

Started by Arantor, October 08, 2019, 08:18:50 AM

Previous topic - Next topic

Arantor

So here's a thing... I've been wondering about the PM system for a while and wondering what future it has. I don't mean the SMF 'threaded email replacement' vs something like XF's 'conversations' (that's mostly a UI and semantics debate), but the whole shooting match.

We're rapidly coming to a place where the expectation isn't just 'if a person receives an abusive message, just report it' any more. We're approaching a place where we as forum owners are expected to protect our users, and that means protecting against dick pics, grooming of children/vulnerable people and so on. The theory is becoming more and more the case that people aren't going to report PMs until possibly too late (if at all; reporting someone who has that kind of hold over you is... complicated)

I've wondered if images shouldn't be embedded automatically (just the link instead) in PMs unless the person explicitly marks the sender as some kind of 'trusted' contact which might make the 'surprise dick pic' phenomenon a little less annoying. (It's also incidentally why I don't support PM attachments.)

But I'm really wondering whether we as site owners can and should do more about this kind of thing... I know of a small number of cases where people have looked at PMs (after disclosing that the site owners can do so at any time as part of the site T&Cs) and have made police reports.

I do also wonder if we need to spend some time as site owners doing something to break the idea that PM means 'private message', which it doesn't and never has; in SMF it has always been 'personal message' and in some respects I think XF's rebranding of it as 'conversations' does something positive in that respect. (Though it confuses users and many site owners rebrand it back to 'Messages')

Would be interested to hear more thoughts.

d3vcho

The solution would be to provide the PM system with a full set of privacy settings. We can gather some of them from social media (Twitter, for example) and include new ones. The system would become kinda complex but we would be winning a lot in privacy.
"Greeting Death as an old friend, they departed this life as equals"

Arantor

Except that, by its nature, privacy is the opposite of what I'm getting at. If you are a user on the wrong end of a psychological manipulation game, you might not report something that you probably should, and improving privacy on that score *does not help you*. It actually enables a predator.

The tools that are available in Facebook and Twitter are only useful *if someone uses them*. That is, unfortunately, not the situation I'm getting at - I'm talking about people who don't feel they *can* report something. What should we be doing about that?

d3vcho

Well... Using TensorFlow to identify dick picks? 8)
"Greeting Death as an old friend, they departed this life as equals"

Arantor

You kid but I've been talking to some of my colleagues who've been using ML with the data sets of people in university interacting with their online learning tools to identify users who are at risk of dropping out. There's even been prototypes of sentiment analysis being done on forum posts that feed into that data set.

The only problem is that you'd need a vast amount of data to train it in a useful way and realistically that's not just a violation of GDPR and of ethical standards, it's also fundamentally flawed in that such accusations if flawed and incomplete can easily be life damaging.

d3vcho

Clearly the future of this is to implement something ML or AI related. As you know, GDPR was meant to make big companies check everything that's going to be posted. If those large companies had the enough resources to do so, it'll all be done with ML and AI but that would require A REALLY HUGE amount of resources that they don't actually have. It shouldn't be too (too) difficult to design a machine that detects NSFW content and censores it.
"Greeting Death as an old friend, they departed this life as equals"

Arantor

No, the GDPR wasn't meant to make big companies check everything that's going to be posted. That really wasn't what it was about but is a common part of the misconception spread mostly by its detractors. It's mostly about making sure that you obtain consent from people whose data you're going to do something with, and that if you're going to make decisions about people based on that data, that you obtain consent for it. (Plus things like making sure you clean up after yourself on request, and letting people see and update their data where relevant.)

It also has nothing to do with the subject at hand, except about whether you can toss all the data into a big ML and make decisions about it like refusing service based on a machine-backed decision without appeal to a human.

The future is also in that place a terrifying place, where systems that literally no-one understands will be making decisions about peoples lives.

Consider the situation I'm actually talking about: two people on a forum, one of whom is grooming the other for (illegal) sexual content. This happens. It has even happened on sites run by various people here, so we can safely discard the 'it doesn't happen' card. Are we *really* going to put faith in an algorithm (which, remember, can't be adequately explained as to how it arrives at a conclusion in a conventional neural network fashion), to decide if a person is actively grooming another person? What about the opposite case of two adults flirting, but then one is accused of sexual harassment on the back of it? You *really* want to trust that to an algorithm? An algorithm that can't be conventionally examined?

As for determining something as NSFW, good luck with that. What defines NSFW? If you're talking about NSFW as nudity, better hope that you're not in, say, the art world where pictures and statues from hundreds of years ago are on show that are publicly visible (even to minors) but aren't somehow NSFW. Michelangelo's David for example.

Or, for more complexity, the amount of people that post baby photos in the bath on Facebook. That, societally, aren't somehow NSFW or inappropriate.

And then you have what happened to the owner of DragonByte Tech - a forum software provider - where he got blasted for having manga that was... somewhere around the legal grey areas.

This really isn't as simple as you're making it sound.

Bloc

I would think utilizing the buddy system more would help in this respect - its IMHO a most neglected feature that could solve issues regarding getting messages from people you don't know. Heaven forbid we should ever think Facebook is something to go by lol..but IMO they at least got the message thing right, that is, a person have to be approved before sending you a message. While that might not make PM as open as before, it would put a fence around potential abuse.

Diego Andrés

Maybe create a 'spam' folder by default to pair with a more robust buddy system, SMF has a label 'system' which I have never used or even read anyone talk about it, could be based off that.

SMF Tricks - Free & Premium Responsive Themes for SMF.

Wellwisher

I know a admin owner of a small forum (4k members & approx 50k posts) who actually disabled his PM system due to some female members being harassed. Most people on the forum were using the PM system genuinely. However the owner decided to punish everyone. The negative affect of which meant the forum began to suffer from inactivity.

I would probably just add a simply Buddy system (whereby members can choose who can send'em dic pics/ PM's. Also devs might want to think about adding defaults options to protect users as soon as they register. Later members can adjust those privacy options to suit them.


Tricky-Ricky

Quote from: Wellwisher on November 11, 2019, 07:15:17 PM
I know a admin owner of a small forum (4k members & approx 50k posts) who actually disabled his PM system due to some female members being harassed. Most people on the forum were using the PM system genuinely. However the owner decided to punish everyone. The negative affect of which meant the forum began to suffer from inactivity.

I would probably just add a simply Buddy system (whereby members can choose who can send'em dic pics/ PM's. Also devs might want to think about adding defaults options to protect users as soon as they register. Later members can adjust those privacy options to suit them.


I totally agree some form of buddy system with the option from its users to accept someone as a friend or simply block that user.

SpacePhoenix

To a degree it's going to depend on the end users of a site, for example an online game you'll get alliance/clan members PM'ing each other, alliance/clan leaders/seniors PM'ing other alliance/clan leaders/seniors. Other sites you might only get PM's between users and admins/moderators

m4z

How about something like the discourse "trust levels" approach? (This doesn't protect very well against abuse by people in positions of "power" (trust), though.)

Hence, combine that with a buddy system for PMs, so that beginning a ("private") conversation is consentual, and add an option to not receive images at all (active by default). Maybe also add a separate trust level for the PMs, so that images may only be posted after an amount of messages by each participant in the thread (25? 50?), and/or an option for each participant to decide when/if the other person is allowed to send images. Then when somebody is finally allowed to send images, don't show them by default; just show a very blurred/pixeled version that has to be clicked to be displayed properly.

Make reporting and (reporting+)blocking easy and obvious.
"Faith is what you have in things that don't exist."
--Homer Simpson

Es gibt hier im Forum ein deutsches Support-Board!

Arantor

Discourse trust levels are actually a somewhat broken concept in that it inherently presumes active = trustworthy. Have they changed it yet so that you don't auto get some moderation powers just by being active and vaguely popular?

m4z

Quote from: Arantor on November 14, 2019, 07:05:53 AM
Discourse trust levels are actually a somewhat broken concept in that it inherently presumes active = trustworthy. Have they changed it yet so that you don't auto get some moderation powers just by being active and vaguely popular?
According to the link above, no. But their powers are somewhat limited (targeted at "correcting" things and fighting spam), and I assume (although the blog post doesn't talk about it) you can keep users from reaching that level (either by adjusting the limits, or by individually restricting users that abuse their powers. (I don't have much practical experience with the trust levels, I just think it's a great basic concept.)
"Faith is what you have in things that don't exist."
--Homer Simpson

Es gibt hier im Forum ein deutsches Support-Board!

Arantor

It's actually really abusable. Also I wouldn't argue that "renaming topics" or "moving topics" is something I'd let people be auto promoted to by default just by being active/popular.

m4z

I see your point, but what else than constant good behavior should be used to build trust? And if they invested all that activity (or scripted something to look like they did) only to then abuse their powers, you can permanently demote them and they have gained nothing. (These kinds of promotions also might make them re-think their initial goal of being destructive, after all, they're a shining pillar of the community now... ::))
But then again, for me this is mostly theoretical; I haven't witnessed much abuse like that. O:)
"Faith is what you have in things that don't exist."
--Homer Simpson

Es gibt hier im Forum ein deutsches Support-Board!

Tricky-Ricky

Look there is a conversations mod on most other formats that works by selecting Friends well people you get to know on your forum ect with the option to remove them as a friend at any time. So that leaves it up to the user who they want to pm or converse with in still a private manner or not. I think with any system its open to abuse but its for the member to decide.

m4z

SMF 2.1 has/will have a buddy system (I don't know if it's in 2.0 too).
"Faith is what you have in things that don't exist."
--Homer Simpson

Es gibt hier im Forum ein deutsches Support-Board!

Arantor

2.0 has a buddy system too and you can restrict PMs to buddies only if you want but it isn't default behaviour and doesn't have UI to make that so, let alone make it mandatory.

Advertisement: