On Immigration Processing

Millions of people are suffering. Vast amounts of money are being spent needlessly. All because of an antiquated and inefficient methods of processing immigrants.

I personally, together with loved ones, have struggled, waited, and lost battles with immigration authorities. Application processes are difficult, time-consuming, and costly — both for the individual as well as for the authorities doing the processing. The long waiting periods, sometimes years, are extremely difficult for those involved. And the outcomes are often illogical and inexplicable.

I therefore propose we start applying modern technologies and techniques to process to reduce time, costs, and suffering by orders of magnitude, while at the same time ensuring complete openness around the decision and absolute correctness of outcome.

The process (today)

An immigration application process is actually quite simple. And it consists of only 2 steps as seen from the processor’s (authority’s) viewpoint:

1. Collect and verify data relevant to the application.
2. Apply all relevant rules and regulations to the collected data.

Efficiency in the first step is of course highly dependent on the applicant’s ability to provide verifiable data. In this part, the applicant has the ability to influence accuracy and speed of the processing by providing required data and ensuring it is verifiable. But it can often also be difficult for the applicant to figure out what data is actually required.

The second step is highly dependent on the processor’s ability to navigate, interpret, and apply all relevant rules correctly in each case. As long as this part is done by humans (clerks), the process will be time-consuming, opaque, and prone to errors and subjectivity.

The result is often a difficult and frustrating process of collecting and submitting documentation, followed by a very long and stressful waiting period. Sometimes multiple rounds of this is required.

A better solution

Let us in stead imagine a system where the applicant goes online, selects a permit, enters some data, and is immediately (in a matter of seconds!) informed of the result of his application.

If the application is approved, it is based on the assumption that the data he entered is correct, and that he is able to provide the required documentation. He is presented with a list of documentation he must provide. He collects the documents and brings them to a point of authority. The authority verifies his documentation, and the the approval is ratified — presumably within days (not months or years).

If the application is denied, the applicant is presented with an exact overview of every part of the decision tree, showing influences, limits, and weightings of each element. This allows the applicant to discover any errors that may have been made on his part, as well as to learn how he might alter his situation to influence the outcome of his application. For example, if his application is denied because his income is to low, he can try to enter a higher income into the system, verify that this will indeed result in an approval, and then do what he needs to do to ensure that he in fact gets this higher income.
Alternatively he might choose to apply on different grounds, and see if this will result in an approval, e.g. he might apply on basis of family reunion in stead of based on employment/income.

Implementation

All that is required is an implementation of a “decision tree” or “expert system”. It doesn’t even require anything as advances as AI or machine learning. Such systems are in use today in a huge variety of industries — ranging from medicine to flight crew management. But for some reason it doesn’t seem to be in use in something as fundamental as immigration application processing.

The many clerks processing applications don’t arrive at a decision based on personal opinions and sentiments. In stead they apply their knowledge of rules and regulations to an application: They know from documentation and experience how a complex network of decision points should be applied; which should be applied, and how they should be weighted, balanced, and influence the outcome. They know what to look at, what to research, and who to talk to find out what they need to know. And yet this process is time-consuming, imprecise, and opaque.

In stead this process could be modeled in computer code:
“If A and B, then YES, else NO”.
“If B and C, then if D then YES, else if E, then YES, else NO.”
“X = 20%, Y = 35%, Z = 10%. If X+Y+Z > 60% then YES, else NO.”
etc.

This is simply applying function calls to data. Implementing such a system then, is actually a matter of implementing a set of functions or “rules”. The rules can have “validity” parameters such as dates attached to them. They can also link to relevant laws, regulations, rulings, and comments and explanations.

It would take possibly only fractions of a second to complete a “calculation” for an application, and one would be able to see both the values that went in to each function, and the value that came out of each function — allowing anyone to examine the balance and interplay between the different elements. And finally, as the system would know what data it needed for a specific function, it would immediately be able to list any and all data required to do a calculation for a certain case.

And again, because every part of the decision is visible, then either errors will be discovered and corrected, or the applicant will make appropriate adjustments to influence the outcome.

The benefits of such a solution, beyond simply shortening the processing times, the costs, and the suffering by orders or magnitude are many:

1. The decision cannot be disputed. Every part of the decision is made visible, with linking to relevant laws and regulations. No part can be attributed to clerical subjectivity or errors. Any errors in interpretation or implementation is quickly discovered and corrected.

2. Variants and time is part of the implementation: A rule may take effect at a certain date. This allows both applicants and rule-makers to test the outcome of an application before and after the rule takes effect. It allow law-makers to test the implications of a proposed change on existing sample applicants. It also allow applicant to discover why an application gave one result in the past, and gives a different result now. Basically it allow one to “run scenarios”.

3. It frees up a huge amount of clerical capacity. In stead of spending inordinate amounts of time processing applications, some of this human brain power could be put to much better use. Some capacity would be spent implementing and coding rules precisely into the system. Some would be applied to researching the outcomes of the decisions and working on ways to improve outcomes. While much of this freed capacity can be applied to other societal issues such as integration, education, health, inequality, innovation, research, etc.

Conclusion

Immigrants are a valuable resource that is being squandered by keeping them waiting and making them suffer. The many clerks processing applications is a valuable resource which is also being squandered. The technology is available to radically improve the situation. So let’s fix it.