Guus Bosman

software engineering director

You are here


A court ruling on the legal meaning of "strictly random"

Yesterday a district court in Washington, D.C. issued its ruling in a case that boiled down to the definition of "strictly random".

Every year, the U.S. government organizes a lottery that allows 50,000 people world-wide to win permanent residency to the States. The goal of the "diversity lottery" is to make green cards available to people from low-immigration countries.

In the 2011 drawing a computer programming error was made and two weeks after the official drawing of the lottery the Department of State closed the website and voided the results:

We regret to inform you that, due to a computer programming problem, the results of the 2012 Diversity Lottery that were previously posted on this website have been voided. They were not valid and were posted in error. The results were not valid because they did not represent a fair, random selection of entrants, as required by U.S. law.

People that had originally learned that they had won a green card, were upset about this and a lawsuit in federal court tried to stop the government from re-doing the lottery. Their injunction sought some form of relief for people who thought they had won but two weeks later saw their hopes go up in smoke.

The key question that judge Jackson answered in her resolution of the case was whether or not the first drawing was indeed "strictly random", as required by law. If it was not random, the results would indeed have been void, just as the State Department said, and re-doing the lottery would have been a reasonable recourse. The statue provides:

Immigrant visa numbers made available under subsection (c) of this statute (relating to diversity immigrants) shall be issued to eligible immigrants strictly in a random order established by the Secretary of State for the fiscal year involved.

Dictionary definition vs scientifically random

During the trial the CIO of the Department explained what went wrong in the selection process. The algorithm that was used only looked at submissions of the first 2 days (instead of during the entire application period) due to a simple programming error. The drawing was thus not random in the strict scientific sense of the word, where each submission had an equal chance of being picked. (An interesting piece of information: at its most busy time, the Department received about 1 application per second. The first website that I was team-lead for years ago had similar performance requirements and was, incidentally, also associated with a lottery).

The plaintiffs argued that the State's selection was random if one employs "a straightforward
dictionary definition". They cited the Merriam-Webster dictionary which defines
"at random" as "without definite aim, direction, rule or method.", and they provided the Court with an Oxford English Dictionary definition: "having no definite aim or purpose; not sent or guided in a particular direction; made, done, occurring, etc. without method or conscious choice, haphazard."

The court made it clear that this dictionary definition is not the "strictly random" that Congress had in mind. Analyzing language from the State Department's regulations, and examples from laws on casinos and the like, the court came out in favor of the State Department's definitions:

But Congress was not using the term in casual conversation. The statutory provision was written to govern a complex numerical selection process – the manner in which a small group of petitioners would be selected from a pool of millions in a computerized drawing of international significance – and so in that context, the term can only be interpreted in the sense of a technical process.
This conclusion is bolstered by the fact that the word random is not used in isolation, but it is part of the phrase “random order,” a term used in probability and statistics and computer
science. 8 U.S.C. §1153(c). The statute is not satisfied by a result that can ultimately be characterized as random; rather, it calls for a process – the random re-ordering of the data.

In conclusion, the Court found that the State Department had acted correctly in voiding the result and scheduling a new drawing:

The Court is sympathetic to the plaintiffs’ plight. While it does not doubt that the emotional impact of the Department’s reversal has been painful and real, and that many of the plaintiffs have compelling reasons to seek to immigrate to the United States, it must take note of the fact that all of the others who submitted timely petitions during the thirty day period also “played by the rules . . . seeking only to pursue their own American dreams.” There are 19 million more stories, from other lottery participants, many of which may be equally or even more compelling, and it is for that reason that Congress determined that every applicant would have an equal chance of winning the right to apply for the visa. The Court cannot order the Department of State to honor a botched process that did not satisfy that regulatory and statutory requirements. Moreover, the Court does not find that it was arbitrary or capricious for the Department to decide to rescind a lottery that did not meet the single most important criterion for a drawing: a random selection.

The new drawing results of the lottery will be available today at noon EST.


Really? The court couldn't just figure out in 10 seconds that a process which excludes applications that are filed more than 2 days after the opening of the lottery is both methodical and quite definite, therefore not fitting the dictionary definition? The mathematical vs dictionary definition argument is a false dichotomy in this case.

Since there was (apparently) no conscious choice to exclude certain applications, the botched selection does fit the dictionary definition. Of course, if the dictionary definition was accepted, then there would be a question as to whether this was a deliberately introduced bug (which would not be random) or not.

I don't know if it's random or not, but this process, taking account only of the two first days of submission, is not fair. Hencefore, if there is any form of justice in United States, the lottery must be voided.

It is like a hockey match which only the two first minutes are taken into consideration, or a 100 meters run decided on 10 first meters.


And the district court is totally right. There is some justice in United States of America.

(Anyway, you've been slashdotted)

Accepting that the lottery may not have been conducted in a strictly random manner, and considering the hardships the reversal created, could the Department of State, based on a broader legal principle such as accountability of the government to follow through on it's guarantees, have been compelled to allow those initially informed that their visas had been granted to immigrate? Would Congress have to be petitioned to allow such an accommodation? Perhaps the lottery could have been run a second time, following the letter of the law, and the first run retroactively considered a separate offering.
I suppose I see this situation as choosing to follow the letter of the law and thus having innocent people harmed by the government's easily rectifiable error. Surely many people made life changing plans in the two weeks between the initial winners being announced and when the error was acknowledged and the offer of visas revoked. NPR focused on one such story.

First, the DoS needs to follow the law, and presumably there is a limit on the visas that can be granted, so an exception would need to be made by Congress. And perhaps Congress should do that, but I do not believe it is within the State Department's discretion to do that.

Second, while there was harm to the people initially notified of visas, there is also some harm (not necessarily to the same extent) who were not eligible to be considered for visas because of the flaw in the selection routine. As the judge correctly pointed out, some of the people not selected (and not eligible for selection using the flawed routine) undoubtedly have more compelling reasons to immigrate than some of those selected.

That all said, I think that DoS should be sure of its results before publishing them, and should fire the people responsible for the initially flawed results.

I also have to ask - should DoS have avoided the controversy entirely by covering up their missteps and standing by the original results, or is it better for them to acknowledge mistakes were made and people were harmed?

Too many times, government agencies try to hide or cover up their actions, and for once, an agency acknowledged its missteps and tried to do the right thing. For that, I have to applaude them.

Another great article on processes that are supposed to be random but aren't is available here:

Recent comments

Recently read

Books I've recently read: