Comments on: TwitBlock spam ratings explained

By: isemann

isemann — Sat, 22 Aug 2009 14:05:45 +0000

Very interesting tool you’ve come up with and something Twitter themselves should really be tackling better.

I reckon the bio info (point number 6) is more important than you think. Also that the user name thing should have less importance – there will be so many lovely ladies who won’t be able to use their real names soon! And as you say is programmatically very hard.

[Insert shameless plug here] I have a score of zero by the way peeps – @isemann

By: ASuggestion

ASuggestion — Sat, 22 Aug 2009 05:09:19 +0000

I agree, the “follow back” algorithm, will have to be fixed some how.
for example, I follow GuyKawasaki, Scobleizer, TheOnion, CNNbrkn, CNN_top, etc.etc.

LOTS of “important” people who I want to know what they are saying, but obviously I’m wise enough to know they couldn’t give two dimes what I have to say.(thus will never follow me back).which is cool.

I don’t use Twitter as a way to “keep up with friends”, I use it as a way to keep up with “The World” I care about (Technology, among other things).

Seems as though this algorithm penalizes those who use Twitter it it’s MOST PURE FORM—ie. A Random Tweet to Friends, but mainly a silent observer listening to those with opinions who have value (but will NEVER “follow back”).

Just my $0.02

By: prgrmr42

prgrmr42 — Sat, 22 Aug 2009 03:32:45 +0000

your random name calculator needs some work. “prgrmr” is a contraction of programmer. removing the vowels and doubled consonants is a standard way of contracting in unix & linux programming. there’s an algorithm for “uncontracting” words back into dictionary words that you should look into.

By: Laure

Laure — Sat, 22 Aug 2009 01:16:43 +0000

Tried TwitBlock on my 2 Twitter accounts, with interesting but weird results:

- I have a spamminess rating of 14 on one of my accounts, I guess because I follow 72 people but only have 22 followers. Like Leighton_Poar above, I mainly use Twitter to keep up with professional & news organizations who rarely follow people back + I block spammers and marketeers every day. Does keeping a close eye on one’s feed/followers, and not having the time/use to enter into a follow me/follow you Twitter race, increase the chances of being listed as a spammer? What’s the rationale here?

- On my other account, a lady who has registered to Twitter to follow me but hasn’t posted anything yet and is not following anyone else is registered as a spammer: quid?

Just some user feedback. Feel free to email if you have questions/comments.

By: Leighton_Poar

Leighton_Poar — Thu, 20 Aug 2009 17:44:09 +0000

I got a score of 27, based only on people not following me back. I mainly use Twitter to keep up with celebrities and news organisations who rarely follow people back. Maybe it would be an idea to take that calculation into account somehow, perhaps by ignoring people I’m following who have Verified Accounts. :-)

By: PhotoJoe

PhotoJoe — Thu, 20 Aug 2009 09:04:01 +0000

1. Ignore factor. — Worthless for people who have protected updates. I deny tons of people to try to follow me. If I wasn’t running protected updates I’d have far more followers than I do now, granted 3/4 would be spam-bots.

By: bob

bob — Mon, 17 Aug 2009 06:27:16 +0000

I find I have to view each new follower and look at who they follow to get a good idea if others are spammers or bots. I am seeing the same pictures but with different names of ones I blocked due to offensive content. This is very time consuming but cuts down alot of new followers that are really spam from other account followers.

By: Tim

Tim — Fri, 14 Aug 2009 23:27:52 +0000

Bare in mind that spam accounts often ’steal’ tweets to appear like real people. So analysing these can be very misleading.

By: monettenom

monettenom — Fri, 14 Aug 2009 22:57:39 +0000

Hey, I am still working on an algorithm which identifies spammer. Check out the prototype http://wayati.com I am using other rules and techniques to identify spammer. Mostly I analyze the structure of the last 20 tweets. Currently I am working on a new version of this algorithm which is more object oriented and makes much more checks. Maybe we can work together to get the best possible tool to fight spammer.

By: Duane Roelands

Duane Roelands — Tue, 04 Aug 2009 02:14:29 +0000

Congratulations on the whitelisting!