Interesting thoughts about DV Lottery 2012 by Raevsky

This are very interesting thoughts, shared by user called Raevsky. He tells us, that the DV Lottery 2012 was valid. Read the rest here:

This is what I initially thought about the whole situation.
———————————————————————————
It was pretty much obvious for
me from the very beginning it is extremely important for DV lottery
program to fight fraud. The congressional testimony http://
judiciary.house.gov/hearings/pdf/Edson04052011.pdf shows the
following examples of fraud
Fraud by applicants includes:

– Pop-up spouses. These are wives or husbands who appear
on the case between the time of initial registration and the time of
visa interview, usually because the successful lottery winner has
sold the privilege of applying for a visa as spouse to the highest
bidder. Sometimes the applicant is a victim of pressure from third
party agents, who will use coercion to force the applicant to claim
a fraudulent marriage so that the agent may sell the benefit of
applying as spouse on the visa application.
– Add-on children. Like spouses, added on after
registration, either by claiming new births, children of a former
marriage now in custody of the
applicant or other excuse. As with spouses, sometimes
applicants are forced to accept these add-on children against their will.

Fraud and abuse by third parties is just as prevalent.

– Just as commonly, unfortunately, the applicant may be extorted
throughout the process. In other words, not just forced to pay for
the release of their initial documents, but forced to pay criminal
gangs in order to be allowed to complete their application. Again,
this problem occurs to a limited extent with other visa categories
as well, but because Diversity Visa applicants do not require
relatives in the US and may be of relatively low education and
employment status, they are particularly vulnerable.
– Nor is this sort of abuse limited to applicants who chose to enter
the Diversity Visa program. Consular officers have also seen
unscrupulous agents enroll groups of people without their knowledge.
In Bangladesh, for example, one agent is reported to have enrolled
an entire phone book so that he could then either extort money from
winning applicants who had never entered the program to begin with
or sell their winning slots to others.

That is why this year in DV-2012 program Department of State (DOS)
worked on additional anti-fraud measures.
http://fpc.state.gov/148253.htm shows the following approach to fraud:
The reason behind those changes is primarily to combat fraud. The
Department wants to protect the interests of legitimate applicants
for the DV program and protect those applicants against fraud and
malfeasance that is sometimes perpetrated against them. And
combating fraud remains the number one goal of the program in terms
of how it’s administered.

The instructions that are already available on dvlottery.state.gov
for the DV-2012 program contain a very strong, very carefully worded
anti-fraud message to would-be applicants that they should take
charge of their own registration, make sure that they retain their
confirmation number, and ensure that that confirmation number
doesn’t fall into the hands of anybody else. And in our
instructions, we also reiterate throughout the very lengthy
information package that if an applicant does not enter the lottery
themselves, actually execute their own application, or at least
personally supervise whoever fills their application out for them,
they are quite likely to become a victim of fraud.

It was also quite obvious DV-2012 program was going to use a
different random selection algorithm than previous DV programs. The
first time the instructions contain reference to more “than 50,000
entries will be selected” instead of 100,000 – http://
travel.state.gov/pdf/DV-2012%20instructions.pdf
Why is that? That is because DOS prepared to use a different way of
selecting entries than before.
Instead of selecting entries independently from the pool as before,
this time DOS decided to select randomly an interval of dates. Those
dates were the dates when applicants submitted their entries during
submission period. The whole submission period was divided to
approximately 6 intervals (the data is provided by the forum http://
http://www.govorimpro.us/showthread.php?t=28753&page=5 – 102 (won) + 75
(not won) entries were submitted on 5-6 of October out of 1000
submitted overall, 102 + 75 is approximately 1/6th of 1000) and the
random algorithm chose one of the intervals, and those from that
time interval were selected as winners. That is perfectly random
from all points of view. Nothing wrong is in what is selection from
certain time periods if the periods themselves are selected
randomly. If you repeat this process many times again and again, you
will see that each entry is selected approximately the same amount
of times – that is a statistical test for randomness. That is what
DOS promised and what it did, from my point of view. I have PhD in
Math and I am a specialist in Monte-Carlo simulation and in computer
programming myself. I could easily write an algorithm like that and
make sure it provides random results and statistical tests for
randomness are fully satisfied.
What does this algorithm give DOS? Two things. Both are anti-fraud measures.
1. Fighting pop-up spouses and add-on children. It is not a
secret in some countries fraudulent companies submit entries of
unmarried persons without their knowledge and later extort money
from them for their wins. What they do here is they submit entries
of unmarried people as married and increase probability to win.
Because if those persons are married, under the old algorithm the
probability is twice as high as if they are not – because they could
apply with 2 entries per family and the whole “family” would get
visas. However, in case they are not married, only one person would
win. Then those persons buy a marriage certificate from local
authorities in corrupt countries. And consulates have to figure out
the marriage is fraudulent, what is not easy. By using new strategy
DOS eliminates advantage of this type of fraudulent marriage,
because there is not going to be any use in it. They would be able
to win both anyway (a lot of married couples won for both husband
and wife independently in DV-2012) and their marriage is not needed
for a visa. That was s smart discovery made by DOS to combat this
type of fraud. The same thing could be done for add-on children.
This “more than 50,000 entries” approach means they did not know
back in September how many entries there would be selected in
DV-2012. Because they did not calculate back then how many selected
persons are going to apply with 1 selected entry or with two. And
they wanted to figure that out later, when people start sending
forms to Kentucky Consular Center.
2. Fighting pay-per-win schema. It is not a secret, a number of
fraudulent companies in some corrupt countries bribed local
authorities several years ago and obtained private information of
residents – their names, family status, date of birth and pictures.
Now they submit their entries with the old photos without their
consent every year for free, and in case they win they extort money
from those selected, up to $10,000 per family as an exchange for
confirmation number, what is required to apply for a diversity
immigrant visa. They do not charge for their work, per entry.
Instead, they charge per win from those selected. The new strategy
of DOS would tremendously increase the risks of those fraudulent
companies, because the variance of income of those companies, who
charge per win, would become significant, and the probability for
those fraudulent companies to go out of business by becoming
bankrupt becomes real. In several years all of them could go
bankrupt just because of that smart strategy. Of course,
mathematical expectation would be still the same, but the risk
increases. That is an effective and cheap way to fight fraud. And it
worked well. What I see on the immigration forums is that that
strategy really worked. One of the users at the forum
forums.immigration.com (Fox25) in the thread http://
forums.immigration.com/showthread.php?323624-DV-2012-was-a-SCAM-!!!
&s=5859d161e9c0c0a1c368e3d936827a90 (already deleted, a copy is available at
http://dvscam.2x2forum.tv/t1-topic ) admits he represents a company
in Ukraine (which is the largest country in terms of fraud in
Europe) that claims his company prepared 252643 entries and had been
submitting them 10000-12000 per day from 7 to 29. First day – on 6th
they were submitting fewer entries (2250) as they were testing.
AND THE WINNERS WERE: 1301 wins on 6th, 0 wins from 7 till 17, 57
wins from 18 to 29. He also admits some number of entries out of the
252643 entry pool were submitted without their consent, with very
old photos (it is illegal to have a photo older than 6 months for
the purpose of submitting the entry), and that they charges much
more than $500 per winner.

Here is another interesting thing. According to the report http://
http://www.reuters.com/article/2011/05/13/us-usa-visa-mistake-idUSTRE74C44C20110513
, nobody was disciplined for the error. Is that really possible?

The truth, as it is clear to me, is there was no computer glitch.
Everything was going according to general plan. Anti-fraud measures
were in-place and working well.

So, why would DOS admit it’s guilt and abandon those perfectly
working anti-fraud measures that were well planned months ago
specifically for DV-2012?

Could that be DOS was afraid of public opinion that did not
understand the algorithms were in fact truly random? Did they think
the courts would step in and suspend the DV program until the
proceedings complete and by that time there will be no time left to
issue visas? That is what I do not know. What is clear to me, the
strategy worked, the algorithms could be written so that they were
really random, and still the results would look exactly like they
look. Or could it be that DOS just wanted to end the DV program as
it is reported by Stephen A Edson in the testimony http://
judiciary.house.gov/hearings/pdf/Edson04052011.pdf
“H.R. 704, the SAFE for America Act will solve the problem of fraud
in the Diversity Visa program in the only way that is likely to
work: by ending it”?
I really do not know what was behind DOS decision to admit the non-
existing guilt, as I see it.

At the same time this type of decision leaves a lot of people
rightfully selected very upset without a valid reason. Their life
dreams were easily destroyed by a very political decision that has
nothing to do with reality. Can we believe DOS abandoned those anti-
fraud measures and left a lot of people upset just because it was
frightened by public opinion?
——————————————————————–
Additional explanations and some clarifications.
If previously they selected entries independently from each other, now they decided to do it differently.
They split the whole submission period into 6 intervals. Those 6 intervals have the same amount of entries each, And then they randomly select the period.
Imagine you have a Roulette where there are 6 places for balls. Those 6 are the time intervals. The first interval (having those 2 days only) has 1/6 of all entries. The data from the forum shows that is true, 1/6 of all entries were submitted on 5th or 6th of October.
They randomly select the time interval. Separately for each region. There is 1/36 probability that 3 largest regions (Africa, Asia and Europe) would come on the same interval.
I could provide a second way of explaining it.
I would use a comparison that is not really appropriate for this case, sorry for this example, but it is easy to understand. Imagine you have a crazy person with a gun.
1. The gun shoots with individual shots. They put the crazy gut onto a rotating platform and spin. When the platform stops the crazy gut shoots. They repeat it 100000 times or so, That is the old algorithm, All 100000 shoots are random and independent.
2. This is now an automatic gun. They also spin the platform, just once, and when it stops, the gut shoots with the automatic weapon 100,000 times, not just once. That is why those 100,000 are close to each other. The results are still random. However, those 100,000 shots will be close to each other. That is the new algorithm. From mathematical point of view it is absolutely correct, and it would work in a casino well.

I think I described previously why this type of algorithm would work as an anti-fraud measure on two fronts.
1. Fighting non-genuine marriages. By having family wins separately they would eliminate the non-genuine marriage fraud, because husband and wife would be able to get visas independently from each other, not using a fake marriage they were put into on paper by a fraudulent company.
2. Tremendously increasing variance of income. That is like de-diversifying in investments. Investment into a particular investment mechanism would be here making a choice when to submit an entry. If the fraudulent company diversities, which means submits on all possible dates, they have about the same return, but lower risk. When it does not diversity, what means invests on particular day, the return is about the same, but the risk is much higher.

So, as I said:
1. This type of algorithm produces truly random results. Even though those results are not independent from each other (correlated).
2. The purpose of the algorithm could be anti-fraud measure, because it works like that.

And this algorithm would in fact produce wins on very specific dates.

Whether they had an error in this algorithm, is not clear – they could have an error implementing this algorithm, as in any algorithm. But if they did not have an error in it, it would produce wins exactly like we saw and it would work as an anti-fraud measure.

—————————————————————–
Now, when I read additional explanations from DOS http://barbados.usembassy.gov/pr05162011a.html I see that my initial thoughts were not correct. I see that because of the following reasons I did not know before.
1. The number of entries selected was between 90,000 and 100,000. That is what DOS admitted. According to my prediction it had to be 30-40% more than for DV-2011 because of family situation to fill the 50,000 quota.
2. It looks like at least 3 largest regions (Africa, Europe and Asia) had all the same “winning” dates of submittal – 5th and 6th of October. Otherwise it was not really possible to have 90% worldwide from those two days to win (the figure that DOS released). Logically, the regions should have been drawn independently from each other. There is still 1/36 probability that would really happen with the type of algorithm described, so that is not impossible, but that is definitely not a high probability.
So what I suspect they might have changed the algorithm because the anti-fraud intentions, but they oversaw at least two those things.