Tech Breh Q&A: How Can We Combat Online Trolls and Hate Speech?

[null]

As it says on the tin, what technical steps can be taken to combat online trolls over and above what is being done now?

Is there a solution out there or is it an intractable or impossible problem (for the time being)?

This is a question for techies who understand the underpinnings of the Internet.

Today Instagram announced ("waffled") the following (spoilered) and it got me thinking:

Spoiler: ..

Instagram FINALLY say they will get tough and remove accounts that send racist messages and stop abusers creating new profiles after sickening abuse sent to football stars including Marcus Rashford, Anthony Martial and Reece James

• Instagram revealed new measures to tackle online abuse after recent attacks
• Company does not use technology to proactively detect content in private chats
• Aim to remove accounts in a bid to reduce abuse people get in direct messages

Instagram have finally announced they will crack down on online a**** by removing accounts that send racist messages and prevent abusers from creating new profiles in the wake of the latest spate of sickening racist attacks on footballers.

Instagram vows to get tough and REMOVE accounts that send racist messages | Daily Mail Online

Facebook content policy manager Fadzai Madzingira

 

[null]

not solvable then? #666-gang still sleeping?





bump

 

[Obreh Winfrey]

We don't consult for free

 

[null]

We don't consult for free

 

[Hood Critic]

This isn't a tech problem, its a people problem, any technical solution is going to have undesired effects on the user base.

 

[null]

This isn't a tech problem, its a people problem, any technical solution is going to have undesired effects on the user base.

not necessarily so if it is a technical system that you can gradually roll out.

@Obreh Winfrey still no ideas?

i'll just leave this thread here. maybe someone might have something to say in the future, as the problem isn't going away.

 

[Obreh Winfrey]

Still no money in my account ?


If identities were inextricably tied to your online access then people would fall back. It's easy to be anonymous and act an a$$hole. The problem with that is, regular people will get swept up in the bullshyt. A troll can sit back and wreak havoc since they an see identities but, if they don't participate, stay anonymous.

If trolls were given 40 lashes, they might start thinking twice . Some people need an attitude adjustment.

 

[null]

Still no money in my account ?


If identities were inextricably tied to your online access then people would fall back. It's easy to be anonymous and act an a$$hole. The problem with that is, regular people will get swept up in the bullshyt. A troll can sit back and wreak havoc since they an see identities but, if they don't participate, stay anonymous.

If trolls were given 40 lashes, they might start thinking twice . Some people need an attitude adjustment.

you are assuming that "an ID" has to be your "real human" ID. in a world of one-way-hashing and other irreversible functions you can have a unique identity which is separate from your own real-world ID. which cannot be reverse-linked to it, but which is uniquely derived from it.

lesson 1: an ID is not you. you can have more than one "ID".

i know more about comp sci than you, so better view this as more of a discussion than a consultation if anything brehs should be paying me but I don't need the money

ideas breh ... not consultation. and definitely not "cannot be done's". if I wanted short-witted/-minded objections i would have posted this in TLR.

i trust you can join the dots based on the consequences of my statement above. the next hurdles are replay-ability and barriers and prevention of pre-materialisation / lookup dictionaries. but like I said at the outset I was looking at an exchange of ideas and on thecoli rumour has it that this place is better than TLR for that.

let's leave the TLR grandstanding somewhere else.

ideas ...

 

[Obreh Winfrey]

you are assuming that "an ID" has to be your "real human" ID. in a world of one-way-hashing and other irreversible functions you can have a unique identity which is separate from your own real-world ID. which cannot be reverse-linked to it, but which is uniquely derived from it.

lesson 1: an ID is not you. you can have more than one "ID".

i know more about comp sci than you, so better view this as more of a discussion than a consultation if anything brehs should be paying me but I don't need the money

ideas breh ... not consultation. and definitely not "cannot be done's". if I wanted short-witted/-minded objections i would have posted this in TLR.

i trust you can join the dots based on the consequences of my statement above. the next hurdles are replay-ability and barriers and prevention of pre-materialisation / lookup dictionaries. but like I said at the outset I was looking at an exchange of ideas and on thecoli rumour has it that this place is better than TLR for that.

let's leave the TLR grandstanding somewhere else.

ideas ...

Breh, you're the only one grandstanding. I kicked off part of the discussion with online access tied to your identity and you come back with knowing more about CS than me? Do you really want a discussion or do you want to flex your knowledge?

 

[null]

Breh, you're the only one grandstanding. I kicked off part of the discussion with online access tied to your identity and you come back with knowing more about CS than me? Do you really want a discussion or do you want to flex your knowledge?

just re-read the above. if you think that "no money in my account" et-al is the way this type of discussion should start then

that's my final word on behaviour or anything outside of technical discussions.

 

[Obreh Winfrey]

just re-read the above. if you think that "no money in my account" et-al is the way this type of discussion should start then

that's my final word on behaviour or anything outside of technical discussions.

Dude you're missing the forest for the trees.

 

[Hood Critic]

not necessarily so if it is a technical system that you can gradually roll out.

• You can't gradually deal with trolling. it's rampant and everywhere.
• You end up effecting your rule abiding userbase because of a small group of people being sh*tty people.
• You never want to institute a solution that has the capacity to become larger than the bigger problem.

For the sake of conversation, what you're looking for is a social point system (yes, like what the Chinese are doing) but the caveat is that you don't want it to look like, be advertised as such or for any of the users to know what it is. So while they are quoting, dapping, repping and negging each other, you're quietly assigning a social score to each user. Those who find themselves in the sh*tty/troll member range, get randomly and gradually shadow banned from certain conversations or from other members even seeing their posts (see Instagram).

 

[null]

• You can't gradually deal with trolling. it's rampant and everywhere.
• You end up effecting your rule abiding userbase because of a small group of people being sh*tty people.
• You never want to institute a solution that has the capacity to become larger than the bigger problem.

For the sake of conversation, what you're looking for is a social point system (yes, like what the Chinese are doing) but the caveat is that you don't want it to look like, be advertised as such or for any of the users to know what it is. So while they are quoting, dapping, repping and negging each other, you're quietly assigning a social score to each user. Those who find themselves in the sh*tty/troll member range, get randomly and gradually shadow banned from certain conversations or from other members even seeing their posts (see Instagram).

might it be useful if you could shadow-ban "all" of their accounts across social media? would that be more of a sanction?
might it be useful if users/sites could apriori filter out certain accounts for not being part of a particular system? sort of like how dns/routing works (with routing and caching)?

IMO the answers to these both are yes and yes.

 

[Hood Critic]

might it be useful if you could shadow-ban "al"l of their accounts across social media? would that be more of a sanction?
might it be useful if users/sites could apriori filter out certain accounts for not being part of a particular system? sort of like how dns/routing works (with routing and caching)?

IMO the answers to these both are yes and yes.

So you're looking to punish, track a troll across the internet rather than within an ecosystem?

Eh, that is a tall order and will get messy considering the youngest child in the house decides he wants to be a troll on Roblox now everyone in the house is considered a troll because IP addressing, ISP, Geofencing, etc. Yeah that will lead to a host of issues.

 

[null]

So you're looking to punish, track a troll across the internet rather than within an ecosystem?

Eh, that is a tall order and will get messy considering the youngest child in the house decides he wants to be a troll on Roblox now everyone in the house is considered a troll because IP addressing, ISP, Geofencing, etc. Yeah that will lead to a host of issues.

Looking to (one way) associate ..

I don't think IP's should be used in any of this for reasons which I explain here:

Spoiler: ...

residential public IP addresses as allocated by an ISP typically can be changed by the ISP at any time. changes usually happen as the router does a DHCP call to obtain an address during start-up but may not happen every time. the change schedule is entirely up to the ISP unless you have been allocated an explicit static address.

the net result of this is that residential public IP addresses are shared between different unrelated households over time.

for example at xfinity;

"All Comcast Internet customers have a dynamic IP address. ... Business-class Internet customers have the option of adding a static IP address — one that does not change."
Static IP address - Xfinity Help and Support Forums - 3285663

secondly, if you are working on a private sub-domain behind an aggregated big IP there is no 1-to-1 correlation between your public IP and you or your internet traffic. by banning the public IP a company could simultaneously inadvertently ban thousands of other people / potential customers.