Suggestions for banning fraud from the site

[Skylimit]

Following yesterdays discussion, the senior members seem to have serious doubt about the integrity of the data ... which is of course the horror scenario

Maybe there is another kind of ranking users. Depending of how much their data can be trusted.

A Woombelist : Having an amount of hits out of proportion
A Serialist : Having a certain percentage (for instance more than 50% of the bills) with more or less consecutive serials.
A Heatofeuroper : someone who enters bills in several counties and where it’s obvious he couldn’t have visited them all
An Errorist : a user who has exceeded a certain percentage of errors between serial and printer codes.

Any hints as to how to prevent site abuse (by I think, mathematical models and statistics) can be posted here

I do not know if there is a model for bills, but there is a model for coins ...
If you know how bills and coins should spread, you can detect all data manipulations, and mark them as suspicious ...

The model I refer to can be found at

There is a theory about the distribution of euro-coins, which uses some heavy maths to explain how coins mix, and how long it takes

http://www.wiskgenoot.nl/eurodiffusie/theorie/eurod.pdf

The model should be valid to a certain extent for bills to ... remark : the 2 euro coins seem to travel faster than the little ones ... so one may assume that 5 euro bills go even faster ... and what about the 500 euro bills ... The report was written in april, and is based on a website, where only the distribution mix was measured

If we assume the model is correct and can be applied to bills, we can calculate the odds for a real hit (at least theoretically ... check out the site : I think there is software available to do this for you)

If we know what we expect, we also know what we don't expect ...
Woombelism : an unusual amount of hits can be quantified

Serialism : is a different problem. a serialm check can be run for specific users which enter suspicious data. The check could search for consecutive numbers, or a pattern in the numbers, which is impossible if we expect a complete randomized input. More advanced cheaters will then probably start randomizing their input, making it impossible to do the check... but over time, their hits will not match the pattern we expect ...

To name one thing : a user claiming to generate his input in Germany must be aware that his bills distribution mix, must match the mix that is specific to Germany in that specific month ... Hey, don't we just have this information at our fingertips ... wasn't that the original purpose of the site ... Randomized numbers will therefore also produce a pattern we do not expect.

Heat of Europe problem : This user must be aware that his input, he claims to generate in countries all over the world must match the distribution mix that is specific to that country (provided enough data exists ...)

Intercontinentals : Users claiming to be from another continent but really are not, should be aware of the timezone ... A user claiming to be from the Westcoast, should be aware not to input his data while average people are sleeping...

Errorist-problem : This site could collect the probability that a user generates an error. We then know what to expect, so we also know what not to expect. Users that generate significantly more errors than the average, can be spotted electronically

There is one problem however : in a normal distribution, there are always extremes. Banning the extremes from the database also equals manipulating the data. The very data we use to find out what pattern we expect ...

Maybe the study group http://www.wiskgenoot.nl/eurodiffusie has some ideas for solving your problem ...

Provided we accept that extremes cannot be banned from the database, because this also corrupts the results, we can still implement the skylimit-tree-concept, which is legendary by now and solves all problems. You can't fake a network ...

Kind regards

[EuroBillTracker]

There is one problem however : in a normal distribution, there are always extremes. Banning the extremes from the database also equals manipulating the data. The very data we use to find out what pattern we expect ...

Yeah, that's what holding me back....

Provided we accept that extremes cannot be banned from the database, because this also corrupts the results, we can still implement the skylimit-tree-concept, which is legendary by now and solves all problems. You can't fake a network ...

hilarious

[thomas]

If we assume the model is correct and can be applied to bills, we can calculate the odds for a real hit (at least theoretically ... check out the site : I think there is software available to do this for you)

How would you identify a "nationality" of a note? The first letter of the serial is "only" relevant with the 15 billion notes of the initial production, or maybe only for the 10 billion notes of the initial circulation. From this year on the ECB uses a pooling system where an NCB produces only one or two denominations (alltogether another 4-5 billion nores) and sends the notes to the countries in need.

And even in the initial phase there were cases of frontloading for e.g. Austrian notes to German banks on Jan. 1st: those N-notes never have seen Austria in their life.

Serialism : is a different problem. a serialm check can be run for specific users which enter suspicious data. The check could search for consecutive numbers, or a pattern in the numbers, which is impossible if we expect a complete randomized input.

What is wrong with a series of serial numbers? Someone at a bank deals with those on a daily basis.

To name one thing : a user claiming to generate his input in Germany must be aware that his bills distribution mix, must match the mix that is specific to Germany in that specific month ...

Why is that? There are not enough data for a given country in a given month to make meaningful assumptions about a distribution mix that is so "typical" that deviations should arouse suspicion.

Thomas

[Skylimit]

Someone dealing with consecutive numbers is an extreme ... I will quote here

There is one problem however : in a normal distribution, there are always extremes. Banning the extremes from the database also equals manipulating the data. The very data we use to find out what pattern we expect ...

Also users producing hunderds of hits do not necessarily cheat (eg one empties the cash machine, the other is in the bank)

Webmaster wrote : Yeah, that's whats holding me back ... So he would rather have a clean database, than allow extremes

Another idea, which is "easy" this time is to give te opportunity to report an abuse ... marking the user "distrusting this user" . If abuse is reported by a large number of users, about the same guy, maybe a clarification can be asked, and eventually further steps can follow ...

[Skylimit]

How would you identify a "nationality" of a note? The first letter of the serial is "only" relevant with the 15 billion notes of the initial production, or maybe only for the 10 billion notes of the initial circulation. From this year on the ECB uses a pooling system where an NCB produces only one or two denominations (alltogether another 4-5 billion nores) and sends the notes to the countries in need.

We can still see who was cheating during the first months, even if we don't know where the notes will be distributed in the future. If a HeatofEuroper was entering 1000 bills spread over Ireland, Portugal, Finland and Greece, in februari 2002, which all happened to start with a Z, then we could say that probably it was a guy from Belgium trying to be funny....

As for the future, we can accept that a bill's nationality is the place where it was first spotted by an EBT-er from this site. We can use the model to calculate the odds, that this very bill, spotted today, in Belgium, will be in Finland within 3 days... possible, but not very likely .... Users who produce this stuff on a daily basis are cheating, or at least not playing by the rules... We can mark them as : suspicious
(user comment : this user has been marked as suspicious by 269 users, for HeatofEuroping)

[EuroBillTracker]

Webmaster wrote : Yeah, that's whats holding me back ... So he would rather have a clean database, than allow extremes

Nope... quite the opposite... I'd rather have the extremes

Another idea, which is "easy" this time is to give te opportunity to report an abuse ... marking the user "distrusting this user" . If abuse is reported by a large number of users, about the same guy, maybe a clarification can be asked, and eventually further steps can follow ...

This is the essential idea behind moderation... there are litteraly hundreds of bills moderated that do not appear in the hit list
I do send emails for clarification but never get an answer!

[bhoeyb]

I do send emails for clarification but never get an answer!

Then, you can (almost) be sure they are fake ... if they are real, you'll answer the email because you don't want to get your special bill hit moderated

[Skylimit]

if they are real, you'll answer the email because you don't want to get your special bill hit moderated

I have the same opinion, provided that you use the language of the user. Not all Germans answer English mails, to name one country. I don't know how Italians, Spanish and Portugease, are with foreign languages.

Also, lots of users give a more or less fake e-mail, because when you register on public sites, you tend to get lots of junk-mail (which is driving me nuts - even with the junk-filter set to exclusive, you still have to go through them if you don't wanna miss anything) So I use my hotmail to register, and occasionally miss a mail or two

[Finn]

I do send emails for clarification but never get an answer!

Btw, did anyone ask about this? What made me suspicious was that "Kalle Laakso" has said about "Very smooth and clean, at least until now", then "DJ LEKA" wrote next day "Still excellent". How did he know what K.L. wrote while he was entering the note, not the mention he's got total one note...

[wouter]

it looks like a false hit to me, although i can't read finnish so it probably should be moderated.

[Olivier]

it looks like a false hit to me, although i can't read finnish so it probably should be moderated.

First of all, all the hits with less than 24h between 2 registrations should be moderated.

O.

[EuroBillTracker]

First of all, all the hits with less than 24h between 2 registrations should be moderated.

What about when people write the URL on the bill ?

[Mr Euro]

I don't moderate bills unless:

a) It's clan entry (pe Woobelism)
b) Accidental entry (pe multiform error)
c) Ask for by the hit person
d) If there are more than one hit by the same persons in a row


I don't moderate bills that are entered on the same day or few days apart because I believe that one bill can be passed on and entered by the another EBT and if there is no relation between those EBT-ers

More Fun!

Mr Euro
<EDITED> I did leave the example out, because it was not valid </EDITED>

[Dakkus]

I do send emails for clarification but never get an answer!

Btw, did anyone ask about this? What made me suspicious was that "Kalle Laakso" has said about "Very smooth and clean, at least until now", then "DJ LEKA" wrote next day "Still excellent". How did he know what K.L. wrote while he was entering the note, not the mention he's got total one note...

Okay. An old thing and so on. But as I don't know how to spot if it's moderated or not, I felt like saying this:
Kalle is a Finnish name. But it's also the name of a Finnish porn magazine. Also laakso is English for "Valley". But think about the name "Hustler Valley" (you know Hustler is a porn magazine, don't you?).. What do you think it actually means (yes, bring your "pervertic" thoughts out).
And then "DJ LEKA". Well. Leka means a big hammer. And also big dick.
So, both names seem to be clear references to sexual inmatureness of the enterer of the notes. Which would also mean higher probability that the person is in the age when faking this kind of info feels ah-so-funny.
Also the post number said to be in Tampere is actually in Helsinki.

So that note is almost 100% surely a fake, as there is very much proof about it's fakeness.