When a court reaches a decision, that will not necessarily mean that the objective of the legal challenge will have been reached. Usually, it is necessary to take additional steps to ensure the decision is executed or enforced in practice.
Those who have taken the case may be lucky, and the decision of the court could be self-implementing. This means that nothing will need to be done to ensure it is actually carried out. For example, the court itself might have the power to invalidate a law or measure that was in violation of human rights. This would mean that, as of the order, the law or measure will no longer have binding effect.
More frequently, however, additional steps will need to be taken to ensure that decisions and orders are actually implemented.
There are some famous examples of human rights cases failing to be implemented years after a court decision had been reached. For example, in 2010, the European Court of Human Rights ruled that a ban on the Pride parade in Moscow, Russia was a violation of the right to freedom of assembly. Two years later, Russia went ahead and banned the event for another hundred years.
This short article will look at the kind of steps a claimant might take to ensure that the case they have brought has an impact and the remedy is carried out.
Reading a decision
To understand what actions will need to be taken after a decision requires a careful reading of the decision itself. Trying to make sense of a decision can often be harder than first expected.
Court decisions vary in format, length and style. Therefore, there is no simple template or formula for reading a court decision. Trying to read a decision can be further complicated by the inclusion of unfamiliar legal terminology and references to legislation and other law.
Reading a decision often involves a careful parsing of the text into a number of components to help separate the important parts from the less important parts.
These are the key things to look out for:
Name/Citation: at the beginning of a decision, you will usually find the parties listed to the case. Including the parties taking the case (i.e. claimants) and parties against whom the case has been brought (i.e. defendants). As well as the details of their legal representatives. There may also be reference to third-party interveners in the case. You are also likely to find a citation at the beginning of the decision, this is the unique reference for the case and helps people find the decision in databases and law texts.
Court: at the beginning of the decision, you will usually find reference to the court that has handed down the decision. This is important information, as decisions can always be appealed (so make sure you have the decision of the highest court to have heard the particular case) and decisions of lower courts often have less precedential value for other courts that are considering similar cases.
Facts: the decision will often set out the factual background to the case. In cases concerning automated systems, there will often be significant gaps in the facts available to the court, with source code, algorithms and similar information not being made available to the court or the parties to the case. This can also lead to disputes on the facts of the case. For example, in the case challenging an automated system that was used in the Netherlands to produce risk reports on individuals flagged as being likely to commit fraud, the parties could not agree on whether the system used involved machine learning algorithms or a “simple decision tree.” Where there is disparity or disagreement on the underlying facts, a decision from the court might also include “findings of fact,” where the court rules on what actually happened in a given case.
Applicable Law: the decision will often set out the relevant law that it will apply to the facts of the case, this might consist of treaty or constitutional provisions, legislation, or case law. Read together, these will be the rules or tests that the court will use to reach its decision. The decision will also set out the legal question or issue that the court is trying to answer through its decision (e.g. “the issue before the court is whether the use of automated facial recognition technology is a justified restriction on the right to respect for private life under Article 8 of the European Convention on Human Rights”).
Arguments: the decision will also include the arguments of the parties. Some courts will include reference to these arguments in amongst their reasoning for reaching a decision (e.g. some of the decisions in the case studies here and here do this). It can be really easy, particularly if you are reading through a decision quickly, to mix up an argument put forward by a party with the reasoning of the court in the case, so be careful to separate them out when reading through the decision.
Reasoning: the decision will also include the court’s analysis, opinion and rationale for reaching the outcome that it has reached. The court’s reasoning can itself be separated into different levels of significance. The most significant or weighty part of the decision will be the core principle or principles that the court has relied on in reaching its decision. This is the essential element of the decision, and it is the fundamental part on which the outcome rests. This part of the decision carries the strongest weight and will often be binding on other (lower) courts. In common law countries, it is referred to by the name ratio decidendi. Other parts of the reasoning will be of less significance but can still have a persuasive impact on the decisions of other courts. For example, if a court remarks on what the outcome might have been in a hypothetical scenario or if they offer an opinion as an aside, this can still be referred to as part of the court’s reasoning (but it will not be binding on other courts). For example, the first case in the UK to challenge the use of automated facial recognition technology by police included a remark by the court that carrying out an assessment of its potential discriminatory impact before or during a trial of the technology “could be said” to be more important than carrying it out after it had been fully deployed. This observation went beyond the rationale for the court in finding that the specific police department, in the specific circumstances of the case, had failed to discharge their duties under equality legislation. Nonetheless, it was still part of the court’s opinion. This part of the reasoning, in common law jurisdictions, is referred to as obiter dicta. It can often be difficult to differentiate these different aspects of a decision, so it is always worth consulting a lawyer to help work out what the essential elements of a decision are and what their legal impact might be.
Orders: the decision will also usually conclude with the orders of the court. Sometimes, however, orders may be handed down separately to the decision itself. As noted on other parts of this site, orders can differ in substance. For example, orders might be for a law or policy to be overturned, for damages to be paid, for one of the parties to do or stop doing something, or a declaration that something has happened.
The last two aspects of the decision will be crucial to the steps that will need to be taken in order to follow up on the case.
Enforcing a court order
When a court makes an order against an individual or entity, it will be legally binding on that party.
An order can be made at different stages of the litigation. Orders that are made during the proceedings are often referred to as interim orders. These usually deal with case management issues, such as requiring a certain party to appear before the court or disclose certain information. They might also be orders requiring a party to do something or stop doing something while the court is considering the case. Orders that are made at the end of proceedings before a court, i.e. after a court has made its final decision on the matter, are usually referred to as final orders.
If a party is refusing to carry out or abide by one of these orders, action can be taken to enforce it.
Since orders are legally binding on relevant parties, it may be possible to bring a party before the court again or call for criminal proceedings to be taken against them for their failure to comply. Even the threat of such action may put enough pressure on the party to carry out what is required of them according to the order. Orders that are of a declaratory nature will not specifically require a party to do something, making them more challenging to enforce (see below).
It is quite common for a party to try to avoid complying with an order. They will usually do this by appealing the order to a higher court, by saying that they are unaware of the order or by otherwise creating confusion. Therefore, it is usually a good idea for the successful party to identify, and communicate the order directly to, the individuals or entities responsible for carrying it out (in some circumstances it may be a requirement for the winning party to deliver, or serve, the order to the individual or entity responsible for carrying it out before that person/entity will be legally bound by it).
Furthermore, the clearer and more precise an order is, the easier it will be to enforce. This is because there will be less leeway for the individual/entity to allege that they are sufficiently complying with it. For example, if an order states that a party “must carry out a human rights audit” in relation to a piece of machine learning technology, that party might allege that they have complied by carrying out a poorly researched, low-quality and light assessment of the technology.
Persistence is usually the key to securing enforcement of an order. It will involve applying sustained pressure on the individual or entity that must comply. It may require the raising of public awareness around the fact that the order has not been complied with, making it harder for those responsible to claim ignorance. It may even require more litigation that seeks to penalise the individual or entity for not complying.
Enforcing other aspects of a decision
Some aspects of the decision may be less easy to enforce. For example, a court may make a declaration rather than give an order for a specific action to be taken by an individual or entity, e.g. the court may declare that the use of a particular automated system by a government body is unlawful rather than order that it no longer be used. They may even broadly recognise that individuals have certain rights in relation to the use of a machine learning system.
Furthermore, the decision itself might be helpful beyond the specifics of a court order. For example, the court may provide guidance on how a law should be interpreted and applied in a given scenario (outside of the facts before it).
Take, for example, the case of Loomis v. Wisconsin. This case involved a challenge to the use of a risk scoring model in the sentencing of an individual, Eric Loomis, who pled guilty to a criminal offence. Even though the court in that case found that the sentencing court did not violate Eric’s Loomis’ rights in taking into account risk scores when sentencing him, it still provided guidance on the steps that should be taken to bring sentencing courts’ attention to the risks posed by risk scoring algorithms. Even though no specific court order was made in this case, by interpreting and applying the law to the facts the court had established legal principles that should be followed in similar cases involving risk scoring algorithms in the future.
Even though there may not be a specific court order to enforce against another party, steps can still be taken to ensure the legal principles established in these decisions are complied with going forward:
Publicity: raising awareness about the court decision through the media and the internet, including through the sharing of a clear, digestible explanation of what the court has actually said and the legal requirements that flow from the decision. Getting out before the other party can be crucial in framing the narrative that, even in the absence of a court order, legal requirements still flow from the decision that the other party must comply with.
Campaigning: campaigns can be mounted to bring about the change to law, policy, or practice that the court has declared unlawful (or reasoned could be unlawful in certain circumstances). If the court has failed to sufficiently safeguard rights in the given context, the campaign may even be to strengthen human rights laws so courts can reach a more fair and just decision in the future. Campaigns could include events, demonstrations, petitions, and other forms of advocacy.
Political Pressure: it may be possible to leverage political processes to influence governments or parliaments on the matter. For example, ministers could be enlisted to ask questions to the government as to why it has failed to fully implement a court decision.
Re-Litigate: where governments, companies or individuals continue to breach the legal principles set out in a court decision, it may be possible to re-litigate the issue on new facts and use the previous decision to support the fresh legal arguments put forward.
This is usually the final step in a litigation process. If you have read all the litigation explainers on this site, why not read the Illustrated Guide to the Machine Learning Process or the case studies!
If you notice anything incorrect or missing from our explanations, please let us know through the contact form!