To ensure that society does not lose faith in judicial reform (trust in courts is the lowest after parties), it was offered an automated distribution of cases between judges. This is a serious innovation. But! In developed democracies, everything can be distributed manually, because Europeans have confidence in the courts. And even with such innovations, we still have questions..

Those who want to “resolve” cases in the courts are here to stay. Here it is worth recalling the “Schemes” investigation about the mysterious illness of 29 judges of the District Administrative Court of Kyiv, which left the computer to select only one candidate, or the schemes analyzed by DEJURE with the filing of many claims, or manipulations with the specialization of judges.

And although such scandals have not been written about recently, you need to understand: the deeper and more effective the judicial reform, the more people will want to use the blind spots in the legislation according to the case distribution algorithm.

And they are.

[see_also ids\u003d"

"

One of the most high-profile cases of recent times is the election by the Supreme Anti-Corruption Court of a preventive measure against ex-People's Deputy Sergei Pashinsky, suspected of illegally taking petroleum products worth almost a billion hryvnia..

Pashinsky was not a pioneer in the court in this case, therefore, the decision on the issue of a preventive measure, according to the law, was transferred to the previously selected composition both in the Supreme Anti-corruption Court and in the Appeals Chamber.

Therefore, it is necessary to analyze how the initial auto-distribution as a whole took place in the Pashinsky case when choosing the investigative judge Natalia Movchan back in 2022, and then how the Appeals Chamber of the High Anticorruption Court chose a panel with judge-reporter Nikolai Glotov when considering the appeal of Pashinsky’s business partner Sergei Tishchenko.

Below in Fig.. 1 shows graphs constructed according to the corresponding protocols, displaying the probability of selecting judges during automatic distribution depending on their load factor.

In both cases, the probability of randomly selecting a judge with the least workload is determined on a different principle from other judges and is 50%. In both cases, it was this judge who was chosen.

When the suspect is an opposition member, the case will always have political overtones. And there will be those who want to believe in political persecution.

[see_also ids\u003d"

But these graphs in no way prove the presence of abuses in the Pashinsky case (! This algorithm is embedded in the automated document management system that this court uses to automatically distribute cases..

Since the start of the work of the Supreme Anticorruption Court and its Appeals Chamber, the judge with the least workload has always been given a 50% probability of selection. Therefore, simply according to the theory of probability, in so many cases this particular judge was chosen when considering cases, and there will be a coincidence in both instances, as happened in the story with Pashinsky, in every fourth case.

What does this threaten

The court staff knows in advance which judge will have a 50% probability of selection when automatically assigning the next case.. And when it is the desired judge, it is not difficult for the person in charge of the Apparatus to coordinate the filing of the case through the Electronic Court at the right time or to enter pre-submitted paper documents into the system. And the management of the courts will also have the same temptation - to influence.

There are simpler options. For example, a judge is not assigned cases 14 days before vacation if its duration is at least two weeks. But the Supreme Anticorruption Court established that these days are taken into account when calculating the coefficient (clause 3 of section II of the Principles for using an automated document flow system). As a result, the first few days after vacation the judge will have a constant probability of selection of 50% when automatically distributing cases, until he catches up with his colleagues in terms of workload. Anyone can take advantage of this, and the court may not even be aware of attempts at manipulation on the part of the applicants.

But “possibility of abuse” does not mean “presence of abuse”.

The specificity of the described scheme is that during its implementation there are no traces, because there is no interference in the operation of the automated system, and based on the distribution results it is theoretically impossible to determine whether there was a leak of insider information and an artificial selection of a judge or whether there was none, because all distributions look.

[see_also ids\u003d"

If, for example, Pashinsky begins to accuse the authorities of fraud in the selection of a judge during the consideration of his case, he will not be able to prove anything, even if something similar took place. But the authorities will never be able to prove the untruthfulness of the accusations, thereby finding themselves guilty for many.

The situation looks even more interesting if we compare different courts..

And here's the rice. 2 graphs were constructed based on data from randomly selected auto-distribution protocols in the notorious District Administrative Court of Kiev on the day (December 12, 2022) before the adoption of the law on its liquidation and the protocol on the same day in the Kiev District Administrative Court, which has so far been handed over the work for.

Absurd. But using the example of the most corrupt court, we can demonstrate what an honest algorithm for determining probabilities should look like.

There were no manipulations at this stage in the scandalous District Administrative Court of Kyiv. This was not necessary - at the end of his work, the choice of judge did not affect the ability to obtain any decision on the order.

And in the Kiev District Administrative Court, which is not hopeless in terms of justice, the choice of a judge affects a lot, so they use the same dubious algorithm as in VAKS.

Although Ukraine is a unitary state according to the Constitution, there is no unity in approaches in the same courts in different regions. For example, of the six economic appellate courts, the probability distribution without anomalies in the Northern (Kyiv) and Northwestern (Rivne) courts is assigned a probability of selection of 30% to the judge with the least load, and 50% in the other four..

Of the seven operating administrative courts of appeal, the normal distribution is in the Second (Kharkov) and Sixth (Kyiv), in the First (Kramatorsk) - automatically 30% probability for the judge with the least workload, in the other four - 50%.

[see_also ids\u003d"

Is there a significant opportunity to influence the selection of judges

Only at first glance it may seem that a 50% probability for an individual judge is not so much.

For example, this makes it possible to significantly increase the efficiency of the schemes that were described by DEJURE. If for the selection of judges they could file identical claims 5–10 times, then the same reliability under the new scheme is ensured by one, maximum two claims. And it won't attract attention anymore.

In the appellate and Supreme courts, the required panel is selected, and not an individual judge, so the probability of someone dropping out of the panel with such an automatic distribution will already be 55–65%, depending on the number of permanent panels with the corresponding specialization.

The effectiveness of influence from the work of the Public Integrity Council can be immediately divided into two, because for influential people the probability of choosing one of the decent judges they do not want will always be half as high. And the greater the success of judicial reform, the greater the role this scheme will play. In court, for several dozen respectable judges, there may only be one dishonest one left, but it will be possible to ensure a 50% probability of choosing him.

Where did this algorithm come from

This is the most interesting thing: he was mysteriously born somewhere inside the judicial system from a blank spot in the legislation. The law does not establish requirements for the principle of determining the likelihood of choosing a particular judge.

The current Regulation on the automated court document flow system, approved by the Council of Judges, simply states that “the selection of a judge using the random number method is carried out in accordance with the judge’s workload coefficient at the time of the automated distribution of the court case,” and there is no clarification on how this random number should be determined.

Well, then, in the absence of any regulatory legal acts, they took and incorporated an algorithm into the programs centrally developed under the leadership of the State Judicial Administration, providing the ability for each court to set in the settings what probability of choosing a judge with the least load they want to get.

Sometimes a meeting of individual judges asserts in their Principles for the Use of an Automated System manipulative norms like the following: “In the field “Probability of distribution... to the judge with the least load,” the percentage of the probability of assigning the case to the judge with the largest interval is set to 50.”. But more often than not, as, for example, in the VAKS, the meeting of judges does not make any decisions on this matter, and the courts simply use the program according to parameters configured for unknown reasons.

But the described mechanism has a chance of turning from illegal into a legal scheme at any moment..

The Council of Judges of Ukraine, by its decision, approved the draft Operating Procedure for the Subsystem for Automated Distribution of Cases as part of the Unified Judicial Information and Telecommunication System. “For the judge who has the least workload (load factor), the probability range is calculated to be no less than 50% of the sum of the probability ranges of all judges participating in the automated distribution,” the document says.

As a consequence, firstly, even if in a particular court the majority consists of respectable judges, they will still not be able to establish fair rules for automatic distribution at their meeting, and there will always remain an effective loophole for manipulating the choice of judge. Secondly, the wording “at least 50%” opens up the possibility of safely incorporating into the distribution algorithm an even greater probability of random selection of an individual judge.

The only way to sweeten such a pill was to take into account, when calculating the workload, only the days when cases were assigned to a particular judge, with the deprivation of the right to assemble judges to change this. Not very fair in terms of the distribution of work between judges, but this at least made it a little more difficult to use the scheme for abuse, because after the judge returned from vacation, he did not have a 50% probability of selection for several days, as was described above using the example of the current.

[see_also ids\u003d"

Next, the project approved by the Council of Judges was sent through the State Judicial Administration to the High Council of Justice, which on January 25, 2022 approved it as changes to the Regulations on the procedure for the functioning of individual subsystems (modules) of the Unified Judicial Information and Telecommunication System.

And a month later, the High Council of Justice, at the request of the Supreme Court, additionally made small but fundamental changes, establishing that when calculating the workload, the number of working days worked by the judge in the current year at the time of the automated distribution of the court case is taken into account, unless otherwise established by a decision of the meeting of judges. That is, by default, the mechanism described above is already provided, the most convenient to use and dangerous for justice mechanism.

But after the full-scale invasion, the implementation of the new procedure for the automatic distribution of cases was postponed several times by the decision of the Chairman of the Supreme Court. And then the newly elected High Council of Justice, by its decision, postponed February 28, 2023 for an indefinite period - “until changes are made to this Regulation and the software of the court’s automated document flow system is finalized”.

Does it make sense to artificially inflate the selection of an individual judge to ensure a fair equal workload for judges

Abuses with probabilities in the automatic distribution of cases can be justified by the need to ensure fairness for the judges themselves - this seems to be necessary to evenly distribute the workload between them.

But, although parliament did not adopt the laws of mathematical statistics, they are still in effect. An even workload for judges can be ensured without abuse, especially if you do not focus strictly on the idea of \u200b\u200bequality during the current calendar year.

If a large number of cases are distributed to judges, the workload will gradually level out. Even in courts where the number of cases throughout the year is not very large, it is enough to take into account the indicators for the last few years.

And the rest will be done by the already existing algorithm, which is responsible for the linear dependence in the above graphs. For example, in the mentioned story with Pashinsky in the HACS Appeals Chamber, even on the linear section of the graph, a judge with a load 1.6 times less has an 8.4 times greater probability of being randomly selected. This approach effectively levels the workload of leaders and outsiders for the rest of the year without creating easy-to-abuse schemes.

The best evidence of this can be obtained by simply comparing the load variations in vessels using different approaches. For example, let’s take the administrative courts of appeal, because there are a lot of cases there, and let’s look at the data for one day, December 1, 2023 (until the end of the year, but when the judges had not yet started going on New Year’s holidays).

In the Sixth Administrative Court of Appeal, where there is a linear distribution without anomalies, the difference between judges with the lowest load factor (3.5204) and the highest (3.5736) is only 1.5%. In the Fifth Administrative Court of Appeal, where the judge with the least workload automatically receives a 50% probability of selection, the difference between the lowest workload (43.8961) and the highest (44.2100) is 0.7%.

There is no need to conduct complex statistical studies to see the obvious: the “50% on the judge with the least workload” approach, which creates significant risks of abuse in the distribution of cases, in terms of its impact on the uniformity of judges’ workloads, is actually no different from fair distribution.