[twitter-dev] Re: Abuse of multiple accounts

[Matt Sanford]

Hi there,

We do have a slew of reports and tools for our abuse team looking  
at blocking, duplicates and some "secret sauce" to find bad accounts.  
I'll pass this on and see if it wasn't caught for some reason or is in  
the process of being handled. As far as sharing our data it via the  
API we have no plans to do that. The issue isn't showing the data to  
friends, it's showing it to enemies. I think the development community  
could probably come up with some cool analysis on this, but so could  
the spammers. If you show your opponent all of your cards they will  
raise the stakes.


Thanks;
 – Matt Sanford / @mzsanford
 Twitter Dev

On May 8, 2009, at 8:00 AM, Nick Arnett wrote:

I knew this would happen... one person with a bunch of accounts has  
managed to spam my social network analysis:


http://www.twurlednews.com/2009/05/08/entrepreneurs-wanted-12/

In this case, it is very obviously the same person, since she is  
using the same picture for every account and only slight variations  
of her real name.


I can detect some of this by seeing real names that correlate to  
multiple identical tweets... Curious if anybody else has thoughts on  
ways to identify this sort of abuse.  Perhaps if the API told us  
what percentage of people block each user?


Just noticed that most of her profiles have the same home page URL,  
so that's a strong clue... and most of her tweets contain the same  
URL.


I'm sure that Twitter's fraud group uses some sort of scoring  
system... any chance that any of that data could be shared in the  
API to help automated systems avoid retweeting spam?


Nick

[twitter-dev] Re: Abuse of multiple accounts

[Nick Arnett]

On Fri, May 8, 2009 at 8:32 AM, Matt Sanford  wrote:

> Hi there,
> We do have a slew of reports and tools for our abuse team looking at
> blocking, duplicates and some "secret sauce" to find bad accounts. I'll pass
> this on and see if it wasn't caught for some reason or is in the process of
> being handled. As far as sharing our data it via the API we have no plans to
> do that. The issue isn't showing the data to friends, it's showing it to
> enemies. I think the development community could probably come up with some
> cool analysis on this, but so could the spammers. If you show your opponent
> all of your cards they will raise the stakes.
>

I certainly understand that, but I was thinking more of a score, rather than
any information about what's behind the score, to use as evidential logic. I
can see why it is safer and easier to just keep it all behind the scenes
until and unless the account is shut down.

Any chance of sharing the percentage of people who have blocked each user?
 That's feedback from your users, after all, and thus somewhat "belongs" to
the community.  (There's probably a huge hole in that argument somewhere,
but I'm not going to think about it).

Nick

[twitter-dev] Re: Abuse of multiple accounts

[Patrick Burrows]

Interesting. Most of her tweets seem to be pretty random and meaningless as
well. (though, I suppose the same could be said for many legitimate people.)

 

I don't imagine this is the sort of account Twitter would pick up on and
ban, either (that was my first thought - just wait for Twitter to ban it.)

 

--

Patrick Burrows

http://Categorical.ly

@Categorically

 

From: twitter-development-talk@googlegroups.com
[mailto:twitter-development-t...@googlegroups.com] On Behalf Of Nick Arnett
Sent: Friday, May 08, 2009 11:01 AM
To: twitter-development-talk@googlegroups.com
Subject: [twitter-dev] Abuse of multiple accounts

 

I knew this would happen... one person with a bunch of accounts has managed
to spam my social network analysis:

 

http://www.twurlednews.com/2009/05/08/entrepreneurs-wanted-12/

 

In this case, it is very obviously the same person, since she is using the
same picture for every account and only slight variations of her real name.

 

I can detect some of this by seeing real names that correlate to multiple
identical tweets... Curious if anybody else has thoughts on ways to identify
this sort of abuse.  Perhaps if the API told us what percentage of people
block each user?

 

Just noticed that most of her profiles have the same home page URL, so
that's a strong clue... and most of her tweets contain the same URL.

 

I'm sure that Twitter's fraud group uses some sort of scoring system... any
chance that any of that data could be shared in the API to help automated
systems avoid retweeting spam?

 

Nick

[twitter-dev] Re: Abuse of multiple accounts

[Doug Williams]

Actually this set of accounts are prime targets to eventually get swept up
by one of our automated spam algorithms. This data (a spam score) isn't made
public in large part because the code that performs the science is separate
from the main twitter.com codebase. Additionally, we don't want to reveal
any secrets on how to circumvent our analysis.

If you feel that someone is a spammer, please dm or @reply @spam (e.g. @spam
@WealthWizz ) to help in The Fight Against
Crime (tm).

Thanks,
Doug
--

Doug Williams
Twitter Platform Support
http://twitter.com/dougw




On Fri, May 8, 2009 at 10:27 AM, Patrick Burrows wrote:

>  Interesting. Most of her tweets seem to be pretty random and meaningless
> as well. (though, I suppose the same could be said for many legitimate
> people.)
>
>
>
> I don’t imagine this is the sort of account Twitter would pick up on and
> ban, either (that was my first thought – just wait for Twitter to ban it.)
>
>
>
> --
>
> Patrick Burrows
>
> http://Categorical.ly
>
> @Categorically
>
>
>
> *From:* twitter-development-talk@googlegroups.com [mailto:
> twitter-development-t...@googlegroups.com] *On Behalf Of *Nick Arnett
> *Sent:* Friday, May 08, 2009 11:01 AM
> *To:* twitter-development-talk@googlegroups.com
> *Subject:* [twitter-dev] Abuse of multiple accounts
>
>
>
> I knew this would happen... one person with a bunch of accounts has managed
> to spam my social network analysis:
>
>
>
> http://www.twurlednews.com/2009/05/08/entrepreneurs-wanted-12/
>
>
>
> In this case, it is very obviously the same person, since she is using the
> same picture for every account and only slight variations of her real name.
>
>
>
> I can detect some of this by seeing real names that correlate to multiple
> identical tweets... Curious if anybody else has thoughts on ways to identify
> this sort of abuse.  Perhaps if the API told us what percentage of people
> block each user?
>
>
>
> Just noticed that most of her profiles have the same home page URL, so
> that's a strong clue... and most of her tweets contain the same URL.
>
>
>
> I'm sure that Twitter's fraud group uses some sort of scoring system... any
> chance that any of that data could be shared in the API to help automated
> systems avoid retweeting spam?
>
>
>
> Nick
>

[twitter-dev] Re: Abuse of multiple accounts

[Nick Arnett]

On Fri, May 8, 2009 at 10:49 AM, Doug Williams  wrote:

> Actually this set of accounts are prime targets to eventually get swept up
> by one of our automated spam algorithms.


That's good to hear.  I'm going to wait and see how often this happens
before I start working on new code to detect it myself.

Nick

[twitter-dev] Re: Abuse of multiple accounts

[Doug Williams]

Nick,
We have a "chief scientist" in house who actually manages all of these
algorithms. It is his job to determine how to spot spam and sketchy users
(his words) through the data. I'm sure you can understand why we cannot
share this part of our secret sauce openly.

Also, if there are isolated incidents of spam or abuse that you want to
report, you can always send an @reply to @dougw and I can take care of them
on my own.

Cheers,
Doug
--

Doug Williams
Twitter Platform Support
http://twitter.com/dougw



On Fri, May 8, 2009 at 11:30 AM, Nick Arnett  wrote:

>
>
> On Fri, May 8, 2009 at 10:49 AM, Doug Williams  wrote:
>
>> Actually this set of accounts are prime targets to eventually get swept up
>> by one of our automated spam algorithms.
>
>
> That's good to hear.  I'm going to wait and see how often this happens
> before I start working on new code to detect it myself.
>
> Nick
>

[twitter-dev] Re: Abuse of multiple accounts

[Nick Arnett]

On Fri, May 8, 2009 at 11:56 AM, Doug Williams  wrote:

> Nick,
> We have a "chief scientist" in house who actually manages all of these
> algorithms. It is his job to determine how to spot spam and sketchy users
> (his words) through the data. I'm sure you can understand why we cannot
> share this part of our secret sauce openly.


Really, I wasn't asking for algorithms... I was hoping for scores, but
that's okay.


>
> Also, if there are isolated incidents of spam or abuse that you want to
> report, you can always send an @reply to @dougw and I can take care of them
> on my own.


As long as they are isolated, I'm not going to worry much about 'em.  ;-)

Given that this was the first one I've noticed in months, it seems that you
guys are doing a good job.

Nick