Analysis of a Hate Raid Bot account & Technical Solutions to Solve this Problem
I'm a Twitch moderator for 3 small Affiliate streamers. I'm also a former tech support escalations agent for multiple companies & have a business degree, & I do free small business consulting to help small streamers reach Affiliate, start their streaming business, & grow their own communities on Twitch, because I think people have potential to work for themselves doing a job they enjoy & that Twitch has a lot of potential to be a platform to help them do it.
It was the community potential that made me spend the last 10 months learning the platform to help others who were just getting started & that's why I'd rather not leave, I'd rather just help Twitch Do Better with practical solutions based on my insight, because bots are a complicated issue with a lot of factors that are hard to manage, & these particular bots are more sophisticated than simple chat spam bots.
So I've evaluated how the hate raid bots operate by studying one that follow/unfollowed my streamer & others that caught my attention while I waited for the report to be actioned & for Twitch to shut it down. These bots are exploiting vulnerabilities within Twitch to keep coming back, hosting Twitch Partners to harvest IPs & spread from their popularity & viewers to find tagged accounts to hate raid.
They hide from the search algorithms so they don't come up under a name search & instead reroute to a dormant, inactive, randomly chosen twitch account they base their names off of. These bots have also already caused at least one small non-affiliate streamer to get banned by Twitch & have to fight to be reinstated, & if Hoss00312 hadn't gone on a twitter rant & linked his twitter on his Twitch page, I wouldn't have known he was actually a victim of the bot problem & not a participant.
Hate raid bot Hoss00312_ gained over 10k followers in the 18 hours I monitored it waiting for my report to be actioned, because there was no category that made sense to report a bot that never chatted & nothing about its page or content was the problem, so I had to go for a vague "something else" category which likely put it at a lower priority.
And streamers don't seem to understand a ban doesn't block anything but chat & these bots in particular un-ban themselves, come back, & follow again & again waiting for the day they use a tag to conduct a hate raid. Banning is incredibly easy & the reporting is complicated & overwhelming particularly to someone who's either shocked or upset or afraid, so streamers haven't been submitting reports for Twitch to actually review the accounts, just muting them from blasting their chat.
Below is the practical back end technical changes that Twitch can make that will resolve a lot of issues, help Twitch streamers & viewers understand & utilize the the reporting process, get rid of at least this type of sophisticated hate raid bot & help prevent & manage others that come up in the future.
1- These bots are hosting Twitch Partners to harvest IPs & hunt for targets who stream with identity tags to hate raid.
Solution- Bring back the ability for streamers & moderators to see who is hosting their page in the Stream Manager & the option to manually remove themselves from being hosted.
Add a pop up "why do you want to be unhosted from this channel?" with options including "Suspected bot account" & make that create a report for Twitch Support to review with a link to block the account. If a 3+ month dormant streamer account is added to an auto host list, trigger an email telling them.
2- People believe enough bans & public outcry will trigger Twitch Support to look into an account which demotivates them to report & block
Solution- Lower the threshold for bans on an account to trigger a report to Twitch to investigate, accounts catching 50+ bans an hour are either a bot or a toxic chat troll.
Rename "ban" to "mute" & give a pop up asking if the account needs to be reported for hate speech/hostility or as a bot so users better understand that they need to report it to Twitch Support for review.
3- The reporting process is a complicated tech tree with too many options. It's easy to ban & make it "not my problem anymore" & difficult to report a problem account to get Twitch to investigate.
Solution- Simplify the reporting process, start with WHAT the problem is, then WHERE the problem is with a text box at the bottom for added notes. Make it easy for people to report bot accounts & let back end keyword algorithms filter it down the tech tree to the right department in Twitch Support for review.
Add an internal flag on accounts that spam-report multiple streamers across multiple days to investigate for potential reporting abuse.
4- Hate raid bots fill chat with hate speech & the only option for streamers to spare their followers & themselves is to clear chat for the whole stream, which impacts their content & ability of other streamers or viewers to report.
Solution- Take away "Clear chat" & replace it with "Blackout chat," add censor boxes to the last 2 minutes of chat & time out all chatters for 5 minutes to give everyone a time to submit reports.
Add a subcategory to "View Recent Raids" to see all chat names that came with the raid & a "Report this Raid" mass report/ban/block option.
5- The hate raid bots don't show up in search & scapegoat legitimate accounts & open multiple gibberish string screen name bot accounts to conduct hate raids.
Solution- Internally flag multiple close variations of a name & random gibberish string names so any reports on those accounts are automatically set at a higher urgency for a review of the chain of accounts.
Check your search priority algorithms to see how bots at 25k+ follows are evading search results.
6- Hate raid bots gain followers & grow extremely fast, sometimes 1k follows an hour.
Solution- Flag accounts that gain follows too fast to be organic growth or human effort for review, add an exception for accounts that actively stream content to account for popularity growth.
Hoy tuve un asalto de alrededor de 40 seguidores bots, esto debe arreglarse, es injusto que los streamers pequeños perdamos nuestras cuentas por cosas que se salen de nuestras manos, solucionen la seguridad de los canales lo antes posible.
This needs to be fixed asap.
Had Hoss bots do a mass following. Was not a good feeling. The reporting, banning and blocking process was hard to do proper while it was happening. Because stress levels were rising I made a mistake with a genuine account/streamer! Twitch, please look at the suggestions above cos Twitch needs to be a safe place.
This needs greater attention. I just ran a stream tonight and was assaulted by 25 follows from different variations of the hoss00312_ bots. This is incredibly disrupting for those of us that have follow alerts on screen and/or in action logs. As well as that, the follows mean the bots are able to find you easily again for future attacks should the people behind it wish.
Something needs to be done even if Twitch won't.
The solutions given above are very good ones, but sadly, we've seen how relaxed Twitch is on reacting to problems like this, which is why many of these ideas may not even see daylight for months to come.
So here's another suggestion, there are numerous examples of specially made and approved bots that are created, managed and maintained by numerous streamers or people that support them for use in stream chat.
Twitch should allow the use of these bot types, with VERY careful moderation to avoid soliciting the creation of further hate bots, that could counter bots like these hoss bots and similar.
An example of how it could counter these types of bots would be to use a filter system that would outright block anything with a certain keyphrase in its name from even connecting to the streamer, nevermind following or chatting, in the same way as a regular block should work.
I've seen firsthand how the person behind these bots seems to be adapting their methods as well, first avoiding moderation by declaring the bots in the streamer's viewer list as already suspended accounts, then, in just the last week, they changed to hiding from the viewer list at all, completely evading moderation with both methods.
And IP bans evidently don't work and it's not difficult to see how they don't, after all, so many of us use VPNs these days. So allowing IP bans as a moderator action wouldn't affect them.
Twitch needs to come up with a solution to it themselves or let streamers and fans come up with solutions themselves, either way, something must be done and if there is any effort happening to counter these attacks, I'm open to helping anyway I can as I'm sure many of us would be, because it affects all of us and we're all getting tired of both the attacks and Twitch's silence on the matter.
Sorry, I ranted. TL;DR, let streamers and fans come up with solutions that will work with the right support from the Twitch API.
Edit: I was right and wrong, Twitch has reportedly sued two users involved in these hate raid bots, but on top of that, there already is an existing bot called commanderroot which has the power to remove followers and add known bots, over 438,000 of them, to your block list, which is a perfect example of what I was ranting about. So it can effectively cancel out these attacks in one fell swoop.
Yes, Twitch is doing something. No, it won't stop this for good. We still need more permanent fixes like this commanderroot bot.
These are exactly the things we’ve been looking for.
These are all great ideas... Which is exactly why Twitch won't do it.
I really think these are great ideas!
Great list of ideas that Twitch definitely should consider.
This is an excellent list of solutions. I love the multiprong approach.
Current tech support agent. All of these sound like Valid solutions.
Great stuff! Brilliant list for ideas Twitch could use to stop these boys!
All of hate raid and random bottting and hate follow neeed stop
This is an absolutely excellent list of actions Twitch should start with, thank you so much.