227 lines
8.8 KiB
Markdown
227 lines
8.8 KiB
Markdown
---
|
|
title: CLRG's Cheating Scandal
|
|
date: 2022-10-04
|
|
---
|
|
|
|
$SPOUSE just stumbled across a PowerPoint file with a bunch of text messages,
|
|
pretty clearly illustrating a massive cheating scandal.
|
|
|
|
I won't post a link to the file,
|
|
I won't post any screen shots,
|
|
and I won't reveal any names,
|
|
because I'm not interested in doxxing people.
|
|
|
|
|
|
Summary
|
|
------------
|
|
|
|
The PowerPoint file consists of text messages between someone who I will call
|
|
Sam and a number of CLRG school owners and adjudicators. Sam's messages
|
|
illustrate what appears to be a years-long, maybe decades-long, racket in which
|
|
adjudicators trade favors in exchange for inflated scores of chosen dancers.
|
|
These favors are either reciprocal, or, in one case, sexual.
|
|
I see weak evidence of money changing hands.
|
|
|
|
In addition, it is made clear that a school owner is having an extramarital
|
|
affair with Sam.
|
|
|
|
|
|
Technical Analysis
|
|
-------------------------
|
|
|
|
The PowerPoint document going around appears to be a series of screen shots from
|
|
the phone of an adjudicator who I will refer to as Sam. Sam appears to have
|
|
gone back through chat history, taken a bunch of screen shots, and emailed them
|
|
to someone whose name I don't recognize. The text of two emails is included in
|
|
the document before the screen shots which must have been attachments.
|
|
|
|
The icons at the top of the screen shots indicate that Sam has an Android
|
|
phone. Android allows you to take screen shots by pressing the power button and
|
|
the volume down button at the same time. It's common for people to hit the
|
|
volume button before power, which results in the phone turning the volume down
|
|
just before the screen shot is taken. You can see evidence of this: the volume
|
|
control appears in some of the screen shots. Someone more familiar with mobile
|
|
phones could tell you the Android version and phone manufacturer.
|
|
|
|
The screen shots are of text messages with WhatsApp and Instagram (thanks to my
|
|
daughter for figuring this one out). In both apps, Sam is writing the text on
|
|
the right side of the screen, and the person Sam is chatting with is on the left
|
|
side. That person's name and their profile icon (usually a photo of their head)
|
|
can be found at the top of the screen.
|
|
|
|
The icons make it fairly clear that these messages all came from one phone. The
|
|
main apps I see are Twitter, Spotify, something that's just a dot, and another
|
|
thing I don't recognize that looks sort of like a stylized four-leafed clover.
|
|
|
|
There isn't really too much more to analyze here,
|
|
from a technical standpoint.
|
|
|
|
My Take
|
|
-----------
|
|
|
|
I feel compelled to highlight a few observations:
|
|
|
|
### Sam is a man
|
|
|
|
Due to the sexual messages,
|
|
specifically one where Sam is called a "good boy",
|
|
and other clues leading us to Sam's real identity,
|
|
my team feels confident that Sam is a man.
|
|
|
|
### Sam might be creeping on kids
|
|
|
|
There's a weird message, from Sam, saying "178 is hot". There's no additional
|
|
context here, other than that 178 is a U11 girl, which means she is 11 years old
|
|
or younger.
|
|
|
|
It's possible this was just poor phrasing and "178 is on fire" would have been
|
|
better.
|
|
|
|
### Sam is involved in an affair
|
|
|
|
In the most memorable part of the document,
|
|
one school owner and Sam exchange messages outlining an extramarital affair.
|
|
It's not clear whether the school owner's spouse is aware of this.
|
|
But it is clear that the school owner likes Sam to have his "ass up".
|
|
|
|
### Sam didn't intend for this to become widespread
|
|
|
|
While I don't know what his motivation was for taking and sending these screen
|
|
shots, I feel sorry for this person. Nothing in this document makes it look like
|
|
Sam was trying to publicize what was going on. And there are a few things that
|
|
Sam surely regrets sending, such as the affair.
|
|
|
|
In fact, there are enough clues in these messages that my team is confident they
|
|
know Sam's real name. If we're right, Sam definitely would not want any of this
|
|
to become public.
|
|
|
|
### This is old
|
|
|
|
The earliest message is from July 2016.
|
|
|
|
### This is deep
|
|
|
|
In an exchange on slide 5, 6 of the 14 judges at All-Irelands 2019 are
|
|
implicated as colluding.
|
|
|
|
In an exchange on slide 3, it is implied that up to 5 adjudicators are not part
|
|
of the racket, or are at least not bribeable by the person who refers to them as
|
|
"the panel from hell". Let's be optimistic and assume that all 5 of these
|
|
adjudicators are not bribeable.
|
|
|
|
Taking both of these together, we can guess that 6 to 9 of the adjudicators at
|
|
All Irelands 2019 were in on this racket.
|
|
|
|
There's a lot of math to get into here: accounting for how the overall
|
|
rankings are computed may make it possible for two colluding adjudicators (16%
|
|
chance) to compromise the entire ranking. But starting out with half the judges
|
|
compromised belies severe problems, no matter how you slice it.
|
|
|
|
### This probably isn't as effective as you might think
|
|
|
|
Most of these messages are just lists of numbers.
|
|
One smart cheater sent a photograph,
|
|
which would be easier to remember.
|
|
The reality is likely that they don't remember all the numbers with 100% accuracy.
|
|
This would result either in fair judging,
|
|
or throwing points at the wrong number.
|
|
|
|
The younger dancers would have the most inaccuracy in cheating.
|
|
For the older competitors, the accuracy of cheating goes up,
|
|
as the adjudicators grow to recognize dancers.
|
|
|
|
Another thing that helps fairness is the number of adjudicators. At smaller
|
|
events, there are 3 on a panel, so about a 5% chance that everybody is
|
|
compromised, if half the judges are bad.
|
|
This is why there are more adjudicators at the higher-placing events.
|
|
|
|
### This could be detected
|
|
|
|
If anybody ever bothered to do analysis of the scores handed out, it would
|
|
be pretty easy to detect the type of cheating that this document seems
|
|
to illustrate. That wouldn't stop the cheating, but it would make it harder.
|
|
|
|
Unfortunately, I don't believe any of the software in use does anything like
|
|
this. And I don't think the data is made available in a way that would make
|
|
independent audits feasible.
|
|
|
|
The data exists to be analyzed and audited, but until there's a lot of pressure
|
|
to do this, I don't see it happening. And as angry as everyone is about things,
|
|
I doubt the result will be auditable data. Doing audits is expensive, and the
|
|
people with lots of money have more incentives to use that money to bribe adjudicators.
|
|
|
|
### These messages are just the tip of the iceberg
|
|
|
|
Let's do some arithmetic!
|
|
|
|
Including Sam,
|
|
there are 14 people implicated by this document.
|
|
|
|
We are just reading Sam's messages: presumably everyone involved is messaging
|
|
everyone else as well. Mathematicians call this a "complete graph". That means
|
|
there are `(14*13)/2 = 91` conversations. Put another way, we are only seeing 15%
|
|
of the racket within this group of 14 people
|
|
|
|
But these appear to mostly be UK people.
|
|
If this racket is larger than the UK,
|
|
you can expect much, much more has been going on that we have yet to learn about.
|
|
And you can bet that it's larger than the UK:
|
|
it is difficult to envsion how something like this could stay that regional,
|
|
with an organization as international as CLRG.
|
|
|
|
This doesn't mean your school is involved.
|
|
But it probably does mean your school has been affected.
|
|
|
|
|
|
Background
|
|
---------------
|
|
|
|
### CLRG and Competitive Irish Dance
|
|
|
|
An Coimisiún le Rincí Gaelacha (CLRG)
|
|
is the largest competitive Irish Dance organization in the world.
|
|
Their main activity,
|
|
as far as I know,
|
|
is running competitions, called "Feisanna",
|
|
which Americans pronounce like "feshes",
|
|
because we can't read Gaelic.
|
|
|
|
These events are adjudicated by certified judges,
|
|
most of whom also own dance schools of their own.
|
|
These judges fly out to the events,
|
|
sit through 8-12 hours of mostly children dancing around on a stage,
|
|
try to write helpful notes in just a few seconds,
|
|
assign a score,
|
|
and then move on to the next round.
|
|
|
|
Periodically, the scores are tabulated by volunteers
|
|
(mostly dads, from my experience),
|
|
entered into one of a handful of online systems just for Irish Dance competitions,
|
|
and awards are presented based on results computed by the systems.
|
|
(None of these systems have auditable source code, to my knowledge,
|
|
but that's a topic for another article.)
|
|
|
|
|
|
### About Me
|
|
|
|
I'm a computer security researcher and educator,
|
|
specializing in forensic investigation.
|
|
That means I show up after the bad thing has happened,
|
|
and build a story about what happened based on gathered evidence.
|
|
The analysis I'm doing here isn't particularly difficult:
|
|
my teenage daughter actually helped me put it together.
|
|
But I am coming at this from a history of doing similar things.
|
|
|
|
My daughter does competitive Irish Dance with CLRG.
|
|
She's been dancing for over 12 years now.
|
|
My involvement has mostly stayed peripheral:
|
|
running sound and electrical at events,
|
|
helping with tabulation,
|
|
and driving everybody home while they sleep.
|
|
|
|
I am not a journalist.
|
|
I'm really looking forward to reading what gets written about this
|
|
by somebody who does investigative journalism for a living.
|
|
But that article isn't out yet,
|
|
so I wrote this one.
|