Virginia Murr Profile picture
Without systemic change within platforms, social media will only get worse. My published bot research:
Charlie Helps FRSA 🖤 Profile picture 1 added to My Authors
14 Oct
Troll networks and Amplification networks are often the same entity. In such cases, judging them as separate platform abuses can do more harm than good.

This is how it works on Twitter.

(Long—but hopefully informative—thread)
A political operator “influencer” (e.g., someone who runs a politicized non-profit) builds a large acct through:

1. Acct stuffing (bots)
2. Organizing a troll network in Twitter DM rooms to “protect” his TL & “fight” the opposition (& sometimes competitors in the same space)

1.Building retweet (RT) DM rooms to amplify messaging (& slowly pulls in real users to “fight the good fight” alongside the bots/trolls)

2.Encouraging people running troll & RT accts to create multiple accts “in case Twitter silences them” (alts=bonus followers)

Read 17 tweets
3 Oct 20
Thread of responses to Twitter's poor PR attempt today re: "threats" & other forms of harassment.

Note that the threats and harassment are mostly accepted as common. Also note, in cases where the accts are reported, Twitter often does nothing.

Read 18 tweets
3 Oct 20
I wasn't joking when I said this kind of thing is rampant on Twitter--and has been for years.

Rape threats - common.
Death threats - common.
Doxxing threats (and actual doxxing) - common.
Physical abuse threats - common.
Countless other types of harassment - common.
One of my "fond" memories of a death threat came in the form of a meme (from a troll acct) that said "people want to kill you."

Twitter found no fault with the tweet.
Then there are accts like this. Pure troll. The acct has been harassing and threatening people since Nov. 2019--and it is still active.
Read 6 tweets
19 Sep 20
As expected, Klyushin is tweeting about Coney-Barrett again (as he did in Jan. 2019).

Let me repeat: He regurgitates US news & pretends the ideas are his--to screw with us.

Klyushin is a bot-troll distributor (for Russian ops--as confirmed in Vol. 5 intel report). Nothing more
Almost verbatim, Klyushin regurgitates US news. One of my faves was his thread "directing" Trump about Afghanistan troop withdrawals--nearly every word of which was taken from an Erik Prince interview published just a few days prior.…
I spent a long time source-tracking bot-troll networks to Klyushin (one of many distributors I tracked)

Let me give a tiny example of why it is stunning that Twitter hasn't suspended his main acct.

These are 0 following/follower fakes pretending to be American. Note the replies ImageImage
Read 6 tweets
8 Sep 20
A quick reminder that since Twitter does not require real locations to be posted on accts, fake locations are common.

Be wary about reports that rely on location data without assessing legitimacy of the accts.

A few examples from the notorious model bot network:

FL is popular. ImageImageImageImage
Pick-a-state, any state.

How about NY .... ImageImageImageImage
Or, let's look at a mid-western state ... how about OH? ImageImageImageImage
Read 5 tweets
2 Sep 20
Soc. media "engagement" numbers can only have meaning if the engagement comes from real & individual accts.

With the sheer magnitude of fake engagement--political & commercial--on soc. media platforms, real engagement can't be determined. Thus, the metric is meaningless at best.
At worst, soc. media "engagement" as a metric is inherently deceptive.

Individuals, political groups, etc. purchase and/or create countless fake engagement. "Look how popular!"

Users & media alike validate the deception when we appeal to non-validated "engagement" as a metric.
Until platforms find a way to:
1. ensure that engagement is *real*
2. prevent (or at least disincentivize) algorithm gaming
3. not duplicate counts (Twitter counts multiple accts run by one person as "individual" accts--see SEC docs)

"Engagement" will continue to be meaningless.
Read 4 tweets
19 Aug 20
Short rant:

For years, platforms have been playing good people, journalists, & organizations.

Under the pretense of "fixing" the platforms, they role out PR initiatives that sound oh-so-good. "No more hate!" "No more violence!" "TRANSPARENCY!!!!"
2/ When, in fact, the platforms are doing the barest minimum to fix their deep, systemic, KNOWN problems.

But, well-intended people who want to believe the platforms' PR schtick get roped in. And they even help them with the facade by joining the PR initiatives.
3/ We need to realize that the platforms:
- KNOW what their problems are
- KNOW (better than anyone) how to fix them.
- Have had YEARS to do so.
- Haven't done the hard work necessary to fix the soc. media plague they've wrought.

This is a choice.
Read 4 tweets
27 Jul 20
Some of the virulently anti-Soros accounts seem to have something in common ... I can't quite place it ...

Ah! It must be the "elevated discourse" Jack and his PR team like to talk about. ImageImageImageImage

Seems totally legit ... 🤦‍♀️ ImageImage
There must be a psychological phenomenon that encompasses Twitter trolls' obsession with penises and blow jobs. Oh, and targeted harassment. ImageImageImageImage
Read 4 tweets
19 Jul 20
A very quick Twitter search for the term "vaccinated" pulls up a ton of anti-vax accounts.

I randomly selected one. Went to the account. And, what do you know, some of the accounts Twitter recommeds as a follow that are "similar" are Q accounts. ImageImageImage
And while NOT all of the acct's in his followers are real ... (cough, cough) ...

A number of them are Q accounts.

Anti-vax and Q nutters are absolutely intermingled. And, yes, Twitter's algorithms encourage it. Image
Read 4 tweets
19 Jul 20
Hashtag manipulation = algorithm gaming

Twitter's algorithms often don't catch even these obvious examples of inauthentic activity.

To reiterate a point I've made a few times before (ha!) ...

Trending hashtags mean nothing. ImageImage

More of the same ... ImageImage

And still more ... ImageImage
Read 4 tweets
21 Feb 20
If you think that Twitter is doing enough to clean up its platform, this thread is for you.

You may want to consider what you find when you study the fake acct patterns manually (while documenting them).

The patterns lead you to things like:
Certain commercial accts from all over the world following row-after-row of the same Russian botnet with English profiles—some with clear, political intent.
When you study the patterns, you’ll also find tens of thousands of dormant bots with Russian profile descriptions & English names—shared along a bot-distribution network of mostly real accts.…
Read 9 tweets
21 Jun 19
Knowing that Russia is still actively attacking us via cyberwar:

I have to wonder HOW Twitter's fancy algorithms haven't removed these fake Russian accts that have English names, an embarrassingly obvious pattern, all following the same acct?

The Pattern
1. Names vs profiles
2. How many words can you make out of the phrase “Russian bot”?
Basically, that’s how these bots are formatted.
The Russians took some pictures, first names, & last names, and randomized them. Scrolling the acct reveals the redundancies.

Redundant pics
More redundant pics
Read 12 tweets