Let's be friends!

It is true that a small team will never be able to monitor all messages. However, whilst the idea of a spam filter might seem attractive based on the efforts that people see with (e.g.) Google Mail, you have to remember that this has taken probably hundreds of person-years of effort to implement and relies on sophisticated systems of artificial intelligence to develop a library of key words to weed out.

Even apart from the development effort, it would probably be quite difficult to weed out the undesirables here, since most of the key words that they would be using would be the same ones that most of us are using throughout this site - it's the *context* of their use that is causing the problems here, whereas it is the mere *appearance* of such words at all in regular e-mails that makes them much easier (but still not easy) to route to the spam folder.

Therefore, the only practical alternative is likely to be a team of moderators. By making it larger then 'small', would stand the best chance of catching the culprits and by having it contain a wide range of people with 'protected characteristics', you stand the best chance of having mods who have empathy with the people being targeted.

Protected characteristics: https://www.gov.uk/discrimination-your-rights
How you can be discriminated against: https://www.gov.uk/discrimination-your-rights/how-you-can-be-discriminated-against

(I've just found out that the previous link I gave had accidentally included a closing parenthesis, so produces a 404. Click )

I also think it seems likely that the idea mentioned elsewhere of restricting the number of PMs/day for the initial period of membership (and ensuring that this was prominently displayed when signing up to deter undesirables) would likely restrict the problem of someone joining then immediately spamming individuals with PMs. 'Decent' new members would be happy to read posts and wait whereas the sort we don't want would likely get bored quickly and stop visiting the site.

There is no way something like Google's spam filter would work here, the main way that operates is that because they have billions of users their systems can see when the same message is sent to thousands of people at once, and auto flag such messages as spam.

Also no-one is suggesting that moderators should read messages before they are delivered, that would be a massive invasion of privacy, almost certainly illegal in the EU and I imagine illegal in the US too. Plus the UMD is a US-based site so UK ideas about protected characteristics, etc, don't apply.

I think the overall consensus so far appears to be:
A block function so someone who feels they are being harassed can simply lock the offending account out of their UMD experience.
Rate limits on personal messages plus controls on who can or can't message someone.
A larger and more diverse set of moderators so there can be faster responses and so users can feel there are mods who represent them, including if possible women and Lgbt+ people, or at least visible allies.

There have been several of my models that have ended up VERY into wam after being introduced to it. I always encourage them to join the community to learn more about it, chat with like-minded people and maybe boost their sales. But almost every time one of them joins, they don't stay long, and even lose interest in modelling for videos. It's pretty sad to me that there are beautiful women willing to cater to our interest...even enjoy doing it...and they can't even be treated decently enough to feel ok about it. They have all said the same things to me. Most of the people on there are cool, but some of them creep me out.

I'm not suggesting that they do ... merely that we should use those as an example of best practice and consider those as an aspiration towards which the moderators should work and, preferably from which group of characteristics the mods should be drawn so as to have empathy with the groups that often suffer the worst problems.

Actually, spam filters aren't that difficult to design or implement. You really only need a couple thousand messages to get one up and running with a reasonable amount of accuracy. From a machine learning perspective, it's a pretty simple task.

If all you're trying to do is to filter out messages that actually *contain* certain keywords, then that's possible.

However, name me a keyword that might appear in an unwanted message to one individual that might not also appear in a wanted message (i.e. used in a different context) to another individual.

Spam filters merely remove the message if any of the keywords appears anywhere in the text. What is needed here is a *context* filter and, since a basic filter can't know what the recipient's wishes are or considers appropriate, it can't know whether the sender's message is wanted or unwanted.

Humans are (currently) much better at decoding those. A system that relies on the recipient choosing to report an inappropriate message to a mod is likely to be far more reliable, as human judgement can be applied.

Modern spam filters are a lot more sophisticated than just looking for set keywords. The general idea is that you can train a computer to be able to assign weights to a collection of thousands of words, each weight explaining how likely or unlikely the words are to appear in a spam message.

If you're curious, these videos give an overview of the specific tool I'm referring to (the Naive Bayes Classifier):

https://www.youtube.com/watch?v=CPqOCI0ahss
https://www.youtube.com/watch?v=IlVINQDk4o8

Or a paper:

http://www.cs.ubbcluj.ro/~gabis/DocDiplome/Bayesian/000539771r.pdf

Though it is very simple conceptually, it actually does a very good job as a spam detector. In fact, this person implemented a tool like this on a set of only 260 emails and was able to obtain a 96% success rate in filtering spam messages:

https://appliedmachinelearning.wordpress.com/2017/01/23/email-spam-filter-python-scikit-learn/

An additional nice characteristic of this method is that it's easy to continuously update it over time as more messages are sent and has a low computing cost. Overall, it's simple, cheap, and easy to implement, with a pretty high efficacy.

Of course, there are more sophisticated methods out there which could potentially have even better results (Online Support Vector Machine methods), but as something which is super simple to implement and understand, Naive Bayes is a great first step.

NERRRRRRRRRRRRRRRRRRDS

One small problem - at no point was spam mentioned. The problem is not marketeers spamming users' inboxes to sell stuff, but guys pestering women for wam sex chat and being creepy. What's needed are block and message restriction functions, so creeps can be stopped from interacting with users who block them and those who chose to still accept messages can rate limit them to, for example, no-one can send a second message till the previous one has been answered. Possibly coupled with a system where users have to pay to get private messaging send ability (free users can reply to messages they receive but not create and send new messages), to defeat the kind of people who'd just try and create dozens of accounts otherwise. Basically put a steel chain round creeps' ability to harass people.

This tool isn't designed for one specific context. It is a general-purpose tool which can fit most situations where text needs to be split into multiple categories.


Thank you Regis, that's very nice of you

Yep - absolutely. Bayesian maths is normally a way forward for these types of selection procedures.

However, that does depend on the keywords that you want to reject differing from the keywords that you want to let through. That, in turn, means that your initial definitions (most of which, as shown in those examples, have basically Boolean outcomes when considered individually) have to be distinct and have to have been successfully decided in the initial table so that the chosen results to allow or reject any given message remain correct.

The problem, as I see it, is not in getting the process to do what you're asking it to do. The problem arises in your initial definition tables and calculations, since those then affect all the subsequent calculations.


In a normal e-mail filter, words such as "Viagra" or "Nigerian Prince" or "Cock" or "Pussy" etc. are pretty unlikely to appear in a legitimate business e-mail, so splitting those into categories and defining them in the initial table to work out the probabilities and weightings is relatively straightforward for normal business and household applications.

I'm still struggling to see how you would define pretty much any of the words that might arise in an *unwanted* PM in UMD as being different from the words that might be used in a *wanted* PM in UMD. That is a function of how those words are used and their intent. Since pretty much any words that appear in one list will appear in the other list, then it's not really possible to produce reliable predictions, since the calculations of the initial outcomes are, themselves, unreliable. You either end up killing almost all wanted PMs or letting through almost all unwanted PMs.

Of course, I'm quite happy to be proved wrong and if you're prepared to do all the necessary coding and MM's prepared to open up his code to allow it to be inserted into the site then the problem's completely solved and you can pick up your halo here.

Meanwhile, in the real world, I think it's highly likely that a panel of moderators is probably going to be the actual outcome for limiting this problem, at least in the short to mid-term future.

I would be actually. Most of this could be pretty easily implemented with some simple code. I'll reach out to MM about it.

How are you going to tell the difference between a really kinky, expletive-filled and objectified message that I send to a long term friend from FetLife who's just joined UMD and asked me (elsewhere) to welcome her with such a message as soon as I see her username appear on the forums, and *exactly the same message* sent by a creep to a random woman who just joined and who unlike my kinky ladyfriend is not into (consensual) degradation and enslavement and to whom the message is harassment?

I do not believe that situation can be resolved by code.

Exactly. That's precisely the point.

Realistically, these are the kinds of questions that would come up in developing a solution. In practice, it would be possible to adjust the system to reduce the probability of false positives to an acceptable level.

Additionally, if the system has a low degree of confidence about a given message, it could always be reviewed by the moderation team.

There is no way you can code a solution to that problem, it's not something code can fix. What we need is site wide blocking abilities and rate limiting on PMs, putting users in control of what they receive and from whom.

Actually, yeah, this is precisely the type of situation that code can fix. Machine learning is designed to solve problems precisely like this.

Now, I'm not saying this is the be all end all solution to this issue, I'm just saying that it could drastically reduce the amount of unsolicited messages that people receive. Obviously, there's no silver bullet, but a simple solution like what I'm proposing could probably clear up the huge majority of the issues people encounter.

Thanks for bringing this up Bob! I've also noticed that it doesn't matter how many times this topic comes up... it keeps happening. It hasn't been happening to me lately because I've gone pretty silent compared to how active I used to be on umd. I want to be clear the reason I'm not active is directly related to this behavior and people making me feel like shit. I openly have kinky convos with some people on here, and I'd like the people that send aggressive, over the line, insulting, dehumanizing messages to understand that those people I get "flirty" with, and joke sexually with, started with polite conversation and treating me like a human being and that's how we are able to have comfortable fun flirty umd friendships with time. I've also met a lot of these people in real life. So there's that.

The messages I would get from complete strangers that can't even say hello, or ask how I'm doing demanding I do certain types of videos, or telling me what they want to do to me in great detail completely put me off. That's not fun. That's not sexy. That doesn't make me feel good about myself etc. Hence why I've backed off. Something that was so fun and special and made me feel light hearted and sexy became a chore, and that made me angry. So I'm now over here quietly taking the jobs I want with people who are kind to me and telling most to fuck off. Anyone that follows my page knows I ALWAYS speak up on these posts involving harassment of women on umd (I know others are harassed too and think it's all fucked, I'm speaking for myself though, as a woman on umd)The fact that zero positive change seems to come from it is disheartening. I'll keep speaking up though because occasionally it will lead to me getting an email from someone that suddenly clicks and sees what I'm saying and changes their ways. Not often, but enough that I don't feel I'm completely wasting my breath. Ha.

To answer Bobs question about did I contact a moderator, the answer is no. The reason I didn't is a much bigger and deep seeded issue involving trust and me having very first hand knowledge of "rape culture," and feeling that going through those steps is pointless and will stress me out when it leads to no results. I'm conditioned to this, as I know many other women are, because the results, sadly, are usually nothing changing. It's easier to just walk away and ignore. Not to mention the fact that if you're receiving multiple messages like this it would be a giant burden to report them all! Why should we have to deal with that? Tell off and then ignore until they go away is the best/easiest option most of the time. If I had to report all those people I would have just clicked delete on my page within my first 6 months here. I've almost done that 4 times by the way, feeling completely disgusted and in wanting to scream "ya know what? Fuck you all! Peace!" Haha! Some kind person always pops up though to remind me why I'm here, and renew my love of WAM. I'm happy we are doing this conversation though, because that obviously doesn't solve anything. Thanks for speaking up Bosumbob, because it's going to be you guys that make the difference. As with anything in life the people with the power, as well as the actual offenders, that need to be the ones to make the change. So y'all work on fixing this or share positions of power with women on here. When I tell a guy he's being an asshole or rude or whatever I'm just labeled "the complaining miserable bitch," and anyone on here that actually knows me knows that doesn't describe me at all. I don't have the power to make any impact on someone that has zero respect for me to begin with. Bottom line I come on umd to have fun, feel sexy, and share my messy exploits, not be the god damn behavior police to men that want to pie me and cum on my tights. Y'all dig?

The ghost ban sounds like a cool idea. Tricky tricky! Also a female mod presence, yyuuuueeessss! I vote for Candy, Lisa, Mrs Bee, and Kittenish! Also, is there a way to see if multiple accounts are coming from one computer? Because I'm almost certain the worst of these are coming from the same people from different names. That's just what I feel based on things being said over the years though.

I really like a lot of the productive discussion above, especially about code-based solutions.


I'm told that things were hella worse 10 years ago. But, apart from the disappearance of some notorious creeps, I haven't seen much improvement over the past 3-4 years.

I can think of a couple reasons why things haven't improved:

(1) One is that the mods have explicitly said they don't want to get involved in regulating the community unless they know they're acting in a way that the community can get behind. That's understandable, but the laissez-faire style has some obvious limits. I very much appreciate the attempt to tackle the issue in this thread.

(2) The other is that the community itself seems to support a pretty limited ethic that is based around 'The Golden Rule', or "do unto others as you'd have them do unto you". That rule implies that a masochist can punish people, because the masochist would be happy to be punished. The very idea of consent -- caring about other peoples' wants/needs/desires/whatevers -- doesn't even enter into consideration. It's this cultural blind spot and we have to evolved past it.

PS: I also want to see kittenish as a mod! That can be my Christmas gift.

Thank you so much for your honest words
Really shocking in a way though

It's clear by now there's a massive problem. Something should be done to stop it
E.g. publish the poster's name immediately and the reason why he's been banned. Such a pillory may sound like a medieval way of punishment. However, this would expose the offender i.e. hit him like a boomerang. Also, this might be an efficient way to get rid of rotten tomatoes in several fora at the same time. It's not unthinkable the same person is acting the same way in other fora, is it?

Anna Lee said: "99.762% positive..." That's still .038 shy of the gold standard of 99.8% !

Yes.

Over on http://www.sploshuk.co.uk/forum/ there is a facility to ban people based on their IP address. If someone is posting bad stuff out under several different names but all from the same computer, that should stop it (until he moves to another location ... but is quite likely to give up as most people will be doing it from their home).

I don't know how feasible it would be for MM to implement that here if it isn't already being done. However, in order for it to happen (if it's currently feasible), the mods would need to know of your suspicions so that they could check out whether you're right.

This reminded me that I meant to add in there that I've also had many experiences with autistic wammers. I feel like many would get caught unfairly into the name and shame mix. When I say MANY, I mean MANY. I was thinking about this and then hopped to something else in my spiel and forgot to write about it. I am completely forgiving and understanding of anyone with disabilities/super abilities that may make conversation awkward or difficult. These are never abusive messages though, in my experience. They have been more of the annoying repetitive nature and then I feel like a total asshole for getting short with people when the pieces fall together finally and a safe space is created to move the conversation forward. So, I definitely believe it's worth taking the time to talk enough to get past "what do you like to get messy with." Haha. Y'all know who you are. Much love!

Although I don't always log on, I do follow these threads.

Yesterday I almost flippantly evoked Godwin's law...

Now with the insults to Lisa and Maria, I'm besides myself. Guess
I've lived in a bubble too long.

I'm a hyper active scatter brain that when upset forgets certain crap and moves on usually. So to be totally honest I'd have had to be writing down screen names and making logs to put the pieces together and share with the powers that be. Like I said in my previous post though, I didn't make a job of being the asshole police. I don't want the job!! There have been two people I called out though that admitted to switching names. So my guess would be my other suspicions probably weren't far off base. I haven't been on or talking much though so hopefully this will be the start of a clean slate and some positive moving forward on umd for me. Look at that... Bosumbob being the one to address this is already making me feel better and is pulling me back in! Oooooo Bob!! Got some slime for me brotha?! I heard Gunge is a fabulous detox when dealing with online toxic masculinity... sexaaayyyy!

I disagree completely, with just about all of that. The scenes people produce for sale are fantasies, and most people can tell the difference between fantasy and reality. Lots of people love Game Of Thrones, that doesn't mean that in the real world they go around slicing people up with swords or crucifying their enemies. The primary rule in any fetish community is "informed consent". Before doing or saying anything to anyone, obtain consent and ensure everyone knows and agrees with what will be said or done. The problem female members here have been having is guys sending unsolicited explicit messages, making constant demands for new material, trying to engage in the wam version of sex-chat under the guise of negotiating a custom that in fact they've no intention of purchasing, etc. Yes, a degree of that may also happen to men but the degree to which women tend to be treated as sex objects by large numbers of men is vastly greater. When we finally achieve genuinely equal societies these issue will either go away or at least equal out between genders, but that's still many years away.

There's no reason at all people can't act with respect towards each other, regardless of what kinks or fantasies they are into. Meanwhile however we need to give all users the tools to protect themselves against being used as objects without consent - a block function and rate limits on PMs, plus accountability for actions.

I feel bad that assholes are why Maria is less active here. I think that there needs to be a middle man between people requesting customs and the models. Kind of like the way Mike runs interference for Jessie. I think that as hard as it may be to implement something, it falls on the shoulders of MM to do something to enforce rules or a code of conduct. As long as these pricks can come here and make accounts to harass people they will keep doing it. No one should have to deal with that level of harassment or be harassed to the point they find it a chore to come online and be in the community. Because of some of the things setup, I hate to suggest this, but if MM is not going to do something about the issue, than maybe another forum with people that will act is the place to go.

Are you seriously suggesting that the women and men on here deserve to be harassed because of the content? If you are, fuck you and the horse you road in on.

In contrast, guided by your example, we know that douchebags are forever.