This really very long and quasi-aca­dem­ic post is based on a speech I gave to MBA stu­dents of the Man­age­ment Devel­op­ment Insti­tute of Singa­pore some­time in 2016. Sub­ject: how do we resolve dis­putes and what bor­ders, geo­graph­ic­al or oth­er­wise, do we cross in doing so? Bor­ders and oth­er­wise, ged­dit, I was talk­ing about dis­pute res­ol­u­tion in cyber­space and algorithms.

Well, it took me a while to reach those bor­ders as I was speak­ing. If you are not inter­ested in how I got there, jump to Dis­pute Res­ol­u­tion and Enforce­ment in Cyber­space right away. 

I spoke without notes but, as it turned out when I got back home, with a high fever. Per­haps that wasn’t help­ful. Any­way, here’s what my speech soun­ded like in my head, based on my tran­script. I made a Ger­man ver­sion of it, which can be found here.

Cross­ing bor­ders in dis­pute res­ol­u­tion. A talk like this, by a law­yer, will evoke the usu­al asso­ci­ations: must be about inter­na­tion­al arbit­ra­tion and inter­na­tion­al courts of law. Yes, it is to some extent. But apart from that, shall we attempt to cross a few bor­ders? Includ­ing some so far away that they might seem imaginary?

Old-School Disputes

Per­haps there is no need to go into detail, at least not at the begin­ning, about what a dis­pute or a dis­agree­ment or an argu­ment is. At least for now, may dis­pute be to us what por­no­graphy was to US Supreme Court Justice Pot­ter Stew­art: we know it when we see it.

Parties to a Dispute and Ways to Resolve It

One can be in a dis­pute with one­self (we could refer to this as ‘hav­ing issues’), or with anoth­er per­son or oth­er per­sons (I shall sim­pli­fy from now on by using either sin­gu­lar or plur­al at times even where both of them may be applicable).

How to resolve a dis­pute? A dis­pute can be resolved either by the dis­put­ing parties alone, or by the dis­put­ing parties and a third party jointly, or by a third party alone.

One Party: Resolving Issues

This means that if someone has issues, he can try to resolve them on his own, per­haps by reflect­ing on them or by med­it­at­ing. Or he can try to resolve them by ask­ing someone else for help, like a par­ent, friend, teach­er, col­league, priest or psy­cho­lo­gist. It is con­ceiv­able that this someone else goes over the head of the one hav­ing the issues, although this would raise autonomy and dig­nity con­cerns. If so, this might be a bor­der that should not be crossed.

Among Disputing Parties: Giving in, Imposing or Settling

If a per­son is in a dis­pute with someone else, resolv­ing the dis­pute can con­sist of one of the parties simply giv­ing in. Motives aside – when this hap­pens, the dis­pute is over. Anoth­er way of resolv­ing a dis­pute with someone else is when one party imposes his will on the oth­er. Admit­tedly, this may not foster the rela­tion­ship between the parties, but the par­tic­u­lar dis­pute of the day is resolved. And then there is yet anoth­er way of resolv­ing a dis­pute between parties without involving a third party, namely when the parties come to terms and settle, usu­ally by way of compromise.

Impartial Third Party

All of the above can be achieved with the help of a third party tak­ing sides with one of the dis­put­ing parties. A third party tak­ing sides marches along­side of one party, either openly like an allied army, or not so openly like a bribed judge. In the nar­row­er sense he is an addi­tion­al team mem­ber of one of the disputants.

The third party I am talk­ing about is a sep­ar­ate one. Sep­ar­ate implies impar­tial. Neither is this third party a dis­putant in his own right. For want of any­thing else, his role is that of a dis­pute resolv­er. The extent of this may vary.

Impartial Third Party Resolving on its Own

A third party under­stood this way can resolve a dis­pute without heed­ing what the dis­put­ing parties think or want. In this event, this seems to imply a cer­tain lack of autonomy of the dis­put­ing parties on the one hand, and a cer­tain assert­ive­ness of the resolv­er on the oth­er. It seems to imply that the resolv­ing third party not only enforces the applic­able rules, but is also the one who makes them. A par­ent con­fis­cat­ing a toy which the chil­dren fight about comes to mind. Like it or not – dis­pute resolved.

Disputing Parties and Third Party Jointly: Resolution by Mediation

Fully account­able parties, how­ever, are not neces­sar­ily sub­ject to rules set by someone else when it comes to their private affairs. This includes deal­ing with dis­putes. Thus, fully account­able parties may choose, for example, that their dis­pute be resolved with the help of an impar­tial third party in such a way that the third party merely facil­it­ates set­tle­ment. This is known as medi­ation and the third party a medi­at­or. Ulti­mately, the rules which the medi­ation is to fol­low are set by the dis­put­ing parties. They may del­eg­ate this determ­in­a­tion to the medi­at­or or to someone else, but even this would trace back to the parties. Such rules may com­prise a lot of things: the law, trade prac­tices, social norms, or simply the mediator’s own per­cep­tion of what is fair. In any case, medi­ation is when the parties talk it out, with the help and under the guid­ance of the medi­at­or. If the parties reach a set­tle­ment this way and abide by it there­after, then the dis­pute is resolved.

Disputing Parties and Third Party Jointly: Resolution by Arbitral Award

A third party may also decide a dis­pute instead of the parties, in a man­ner which is bind­ing on the parties. This works to the extent that the parties ‘own’ their dis­pute and thus may choose not only wheth­er, but also how they want their dis­pute to be resolved. The parties ‘own’ their dis­pute to the extent that it affects them alone and no one else includ­ing the pub­lic, which is not always obvi­ous. But if a dis­pute is private indeed and the parties decide to have their dis­pute resolved bind­ingly by a decision maker of their own choos­ing, then this is arbit­ra­tion and such a decision maker an arbit­rat­or. As with medi­ation, the rules accord­ing to which the dis­pute shall be resolved are set by the dis­put­ing parties, either dir­ectly or indir­ectly, for example by ref­er­ence to the rules of an arbit­ral insti­tu­tion. And like­wise these rules may com­prise the law, trade prac­tices, social norms, the arbitrator’s per­cep­tion of what is fair or a wild mix of all that. If the arbit­rat­or finds a decision this way and all parties abide by it there­after, then the dis­pute is resolved. If one of the parties does not agree with the decision or abide by it, it is still bound by it. Because pre­vi­ously it had agreed that it would be. 

Disputing Parties and Third Party Jointly: Resolution by Judgment

Decision-mak­ing by a non-party and a decision which is bind­ing on the parties – sounds a lot like judge and court? Quite sim­il­ar indeed, but the dif­fer­ences are cru­cial. For one, a judge usu­ally has the power, even the duty, not only to con­sider the interests of the parties, but also the interests of the non-involved, namely the pub­lic. This is judi­cial dis­pute res­ol­u­tion, or lit­ig­a­tion. In court the parties do not ‘own’ the dis­pute in full any­more. The rules applic­able – at least in eponym­ous rule-of-law sys­tems – are determ­ined by law. Trade prac­tices, social norms and per­cep­tions of equity may play a part, but only to the extent provided for by law. If the judge finds a decision this way and all parties abide by it there­after, the dis­pute is resolved. If one of the parties does not abide by the decision, it is still bind­ing on it. Not neces­sar­ily because pre­vi­ously the parties had agreed to be bound by it, but because the over­rid­ing rules – the law, mostly – decree it.

There is anoth­er prin­cip­al dif­fer­ence between judi­cial dis­pute res­ol­u­tion and any oth­er kind of dis­pute res­ol­u­tion. It has to do with: enforce­ment.

Enforcement by States: the Executive

The res­ult of suc­cess­ful medi­ation is an agree­ment to settle. Parties with an open mind may be more prone to set­tling than to hav­ing the dis­pute resolved by an arbit­rat­or or a judge. One of the reas­ons for this is that parties which enter medi­ation are still on speak­ing terms in the first place. But anoth­er reas­on is that the parties and their medi­at­or are free to focus on their needs and wants, without hav­ing to deal so much with pro­ced­ur­al issues. Medi­ation is flex­ible; the medi­at­or can, and should, feel it out. On the down­side, this means that the res­ult even of suc­cess­ful medi­ation is noth­ing more than an agree­ment which a party may then choose not to hon­our. If this hap­pens, the parties are more or less back to square one, the only dif­fer­ence being that now they are in dis­pute over one of them breach­ing the set­tle­ment agree­ment that was sup­posed to end the ori­gin­al dispute.

The res­ult of suc­cess­ful arbit­ra­tion is an arbit­ral award, which is bind­ing on the parties because they had agreed that it would be, regard­less what they think of it now. It may be more dif­fi­cult to obtain an award because the parties and their arbit­rat­or are not free to deal with the sub­stance of the dis­pute alone. They also have to con­sider how the dis­pute shall be solved. This is why, as an unfor­tu­nate side effect, parties in an arbit­ra­tion often extend their dis­pute to pro­ced­ur­al mat­ters. Some­times they even sab­ot­age the pro­cess. If a party chose not to hon­our the award, this would be a breach of the agree­ment to arbit­rate. Without fur­ther ado, the parties would be back to square one, the only dif­fer­ence being that now they are in dis­pute over one of them dis­hon­our­ing the award based on the arbit­ra­tion agree­ment which was sup­posed to end the ori­gin­al dispute.

Luck­ily, it is not so point­less. A set­tle­ment (medi­ated or not) which is breached or an arbit­ral award which is not hon­oured, in fact any agree­ment which is breached, can be asser­ted. We meet again one of the ways to resolve a dis­pute – by impos­ing one’s will on the oth­er –, but this time as a way to assert that which has been agreed upon. As to the meas­ures one can think of any­thing under the sun and pretty much all of that has been tried in the course of human civil­isa­tion, from per­sua­sion to viol­ence and everything in between. If per­sua­sion works – fine. But what if one needs to get tough with a party in breach? Then there may be the need to enforce that which is bind­ing. For a long time this meant the law of the jungle, but today the legit­im­ate use of force is usu­ally mono­pol­ised by the state. It has a depart­ment for that: the executive.

The exec­ut­ive is the sis­ter of the judi­ciary, the two of three sib­lings which con­sti­tute state power. This is why judi­cial dis­pute res­ol­u­tion is not only bind­ing on the parties, at least with­in the jur­is­dic­tion of the court, it is also rel­at­ively easy to enforce against the parties (even against non-involved third parties as the case may be), at least with­in the reach of com­pet­ence of the exec­ut­ive. Where the legit­im­ate use of force is mono­pol­ised by the state, the state wants to determ­ine what legit­im­ate dis­pute res­ol­u­tion is. No prob­lem, usu­ally, where a dis­pute is resolved by the state itself. In this case, the exec­ut­ive takes a brief look at wheth­er it is a judi­ciary decision indeed and if so, gets down to action. Medi­ated set­tle­ments or arbit­ral awards on the oth­er hand are obtained without the state; they do not run in the fam­ily, so to speak. To enforce them, one must obtain the state’s more detailed approv­al. In decid­ing wheth­er to grant that, a state would make no dif­fer­ence between them or and any oth­er agree­ment which one party seeks to enforce against anoth­er. All this means that to enforce, one needs a court judg­ment in any case. An excep­tion usu­ally applies, though, for arbit­ral awards.

An arbit­ral award is a decision by an arbit­rat­or. The parties may not be happy with it at all. But they have agreed in the arbit­ra­tion agree­ment that the award, whatever it looks like, would be bind­ing on them. Many, if not most states respect the arbit­ral work done – it does reduce the work­load of their judi­ciary after all – and apply a reduced stand­ard of review to arbit­ral awards in their enforce­ment pro­ceed­ings. They do not address the dis­pute and its res­ol­u­tion as such. They just check wheth­er the award really only affects the parties because they fully ‘own’ the dis­pute, or wheth­er there are any over­rid­ing prin­ciples or interests of non-involved third parties that speak against the enforce­ment of a decision by a non-mem­ber of the judiciary.

Enforcement beyond States: Judicial Cooperation and Mutual Legal Assistance

It lies in the nature of the three state powers – the rule-mak­ing legis­lat­ive, the rule-inter­pret­ing judi­ciary and the rule-enfor­cing exec­ut­ive – that they are con­fined to their state. And ‘state’… well, there are many defin­i­tions of state as a polity, but all of them have one thing in com­mon: they require state ter­rit­ory. This is why usu­ally the enforce­ment of a judi­cial decision is not a prob­lem with­in the ter­rit­ory of one and the same state. But enforce­ment of any dis­pute-resolv­ing decision reaches its lim­its when it has to cross state bor­ders. In the era of glob­al­isa­tion, this hap­pens really often.

To some extent states try to solve this by resort­ing to supra­na­tion­al and inter­gov­ern­ment­al organ­isa­tions and policies. They allow oth­er state powers, or the powers of a third entity, in part to oper­ate on their ter­rit­ory: renun­ci­ation of sov­er­eignty. True samples here­of are the European Court of Justice, the former EU policy of Police and Judi­cial Cooper­a­tion in Crim­in­al Mat­ters, or mutu­al leg­al assist­ance (as far as the exec­ut­ive is con­cerned) and judi­cial cooper­a­tion (relat­ing to the judi­ciary) among the mem­bers of the United Nations, Inter­pol, or the Com­mon­wealth of Nations.

Some­times this works quite well. Renun­ci­ation of sov­er­eignty and inter­gov­ern­ment­al cooper­a­tion may evolve into some­thing rel­at­ively well-aligned, for example at EU level where, argu­ably, the degree of enforce­ment of decisions of the EU judi­ciary by the exec­ut­ive powers of the EU mem­ber states is unpre­ced­en­ted in his­tory. The same applies, with some reser­va­tions, to the enforce­ment of decisions of the court of any EU mem­ber state by any oth­er EU mem­ber state.

In the con­text of dis­pute res­ol­u­tion, there is an even bet­ter example: the Con­ven­tion on the Recog­ni­tion and Enforce­ment of For­eign Arbit­ral Awards. By this New York Con­ven­tion, all rel­ev­ant states in the world have com­mit­ted to do that which its name implies: to recog­nise and enforce for­eign arbit­ral awards – sub­ject, just as a domest­ic award would be, to the ‘usu­al’ checks wheth­er there are any over­rid­ing prin­ciples or third-party interests that speak against the enforce­ment of a decision by a non-mem­ber of the judi­ciary. Or, sim­il­ar to the New York Con­ven­tion, the UNCITRAL Mod­el Law on Inter­na­tion­al Com­mer­cial Arbit­ra­tion, which con­tains pro­vi­sions on the recog­ni­tion and enforce­ment of awards in the field of inter­na­tion­al com­merce. Just that it is not bind­ing and few­er states have enacted it into domest­ic law.

Dispute Resolution and Enforcement in Cyberspace

State bor­ders are the bor­ders every­body thinks of imme­di­ately when ‘cross­ing bor­ders’ comes up. Under­stand­able – we are all spa­tial beings. And we are spa­tial beings because we are phys­ic­al beings. Most of us, how­ever, are more than hard­ware, most of us are also soft­ware. But for so long as we occupy our phys­ic­al bod­ies, these bod­ies have to be some­where in phys­ic­al space. The phys­ic­al space avail­able to us at this point in time is our plan­et Earth.

And because we are many on this plan­et, we live in soci­et­ies, and due to cer­tain bene­fits that come of it many of those soci­et­ies are con­sti­tuted as states, and most states exist­ing today refer to them­selves as rule-of-law sys­tems. So far as the laws of these states con­cern the rela­tions among equal sub­jects, they are called private law. To the extent that the laws con­cern the rela­tions between indi­vidu­als and the state (be it with­in or across phys­ic­al state bor­ders), they are called pub­lic law. Thus, law on the enforce­ment of a dis­pute res­ol­u­tion – executed against indi­vidu­als by state author­ity – is pub­lic law too.

But only a short while ago, from an evol­u­tion­ary point of view, we have cre­ated and star­ted to inhab­it a space incor­por­eal where, as mat­ters stand, there is no pub­lic law includ­ing law on the enforce­ment of dis­pute res­ol­u­tions. Nev­er­the­less, parties have dis­putes there, which are resolved, and these dis­pute res­ol­u­tions do get enforced: cyber­space.

Phys­ic­al space is, well, phys­ic­al, where­as cyber­space is not. We have star­ted to inhab­it cyber­space with the soft­ware part of what makes us up, but we have not stopped inhab­it­ing phys­ic­al space with our phys­ic­al bod­ies. We just do both. There is a fairly dis­tinct bor­der between these two spaces, yet many of us do not real­ise it well, per­haps because it runs right through us and we cross it con­stantly. As a con­sequence, we tend to oblit­er­ate the dif­fer­ences between both spaces. States in the phys­ic­al world have always been seek­ing to reg­u­late beha­viour in cyber­space alright, but come to think of it what they have really been doing is reg­u­late beha­viour in their ter­rit­ory, by way of law, for example on data pro­tec­tion or intel­lec­tu­al prop­erty. It is an indir­ect approach at best. Enact­ing laws and enfor­cing them in the phys­ic­al world may not be dif­fi­cult. Enfor­cing them in cyber­space alone is. There is no dir­ect reg­u­la­tion by law by states in cyber­space, no pub­lic cyber enforce­ment law, simply because there are no cyber states.

How are dis­putes among parties resolved in cyber­space and dis­pute res­ol­u­tions enforced there? I mean, really only there, not indir­ectly by way of tak­ing a detour via phys­ic­al space?

The Law of the Strongest

At the begin­ning we have seen that a dis­pute among two parties or more can be resolved in one of the fol­low­ing ways.

Firstly, one side either wins through, or it gives in. Secondly, the mum shows up and takes away the toy in dis­pute. None of this requires fur­ther enforcement.

Thirdly, the parties find a set­tle­ment on their own, or the dis­pute is resolved jointly with an impar­tial third party – the lat­ter being a medi­at­or, an arbit­rat­or or a judge. These ways of resolv­ing a dis­pute may have to be enforced and thus rely on an enfor­cing authority.

Indeed it would seem that dis­put­ing parties, for want of offi­cial enfor­cing author­ity in cyber­space, are not incentiv­ised to seek a means of dis­pute res­ol­u­tion which might have to be enforced. Instead, dis­putants seem to be incentiv­ised either to impose their will on the oth­er, if they are strong enough, or to give in, if they are too weak. At the same time, they might want to be on the lookout against a ‘cyber mum’ who may appear and resolve the dis­pute the way she deems it best.

What I have just described in the abstract is the beha­viour dis­played by rel­at­ively strong play­ers in cyber­space on the one hand, like Amazon, Face­book, Google or your subred­dit mod­er­at­or, and the rest of us on the oth­er hand. The former tend to impose their will on oth­ers in a dis­pute. And should oth­ers quar­rel ‘under their roof’, they can end this dis­pute either way. All the while, they do not have to heed rules that might be man­dat­ory else­where, such as the rules of pub­lic law in phys­ic­al space, because there is no state mono­poly on the legit­im­ate use of ‘force’ in cyber­space. The rest of us tend to bear the wrong and to give in.

To para­phrase, cyber­space is gov­erned by the law of the strongest.

Example? The data pro­tec­tion com­mis­sion­er of Ham­burg, a fed­er­ate state of the Fed­er­al Repub­lic of Ger­many whose ter­rit­ory cov­ers 0.00051 per cent of the land sur­face area on this plan­et, has ordered Face­book to stop col­lect­ing data on What­s­App users in Ger­many, and to delete all inform­a­tion already for­war­ded from What­s­App on roughly 35 mil­lion Ger­man users. In doing so he invoked powers ves­ted in him by Fed­er­al Ger­man and Ham­burg data pro­tec­tion law. He is now hav­ing a dis­pute with Face­book about Facebook’s beha­viour in cyber­space, where he has no author­ity and where there is no offi­cial exec­ut­ive. Regard­less of what he will try to do about it in phys­ic­al space, it feels safe to say that Face­book will not do as it was told by the data pro­tec­tion com­mis­sion­er of Ham­burg. In cyber­space, Face­book will resolve the dis­pute by impos­ing its will on the data pro­tec­tion com­mis­sion­er. Because it can.

Accord­ing to latest news, the com­mis­sion­er advises against using What­s­App. Put anoth­er way, he advises against enter­ing those regions in cyber­space which are dom­in­ated by Face­book. Sounds like he was giv­ing in.

Private Law

The lack of offi­cial power to help enforce dis­pute res­ol­u­tion in cyber­space does not mean that the pre­val­ent ways of resolv­ing dis­putes and enforce­ment in cyber­space are arbit­rary. The law of the strongest is not applied by guess and by gosh. On the con­trary, we might find that the parties con­cerned fol­low detailed rules on res­ol­u­tion and enfor­cing. Just that these rules are imposed by one party on the other.

Face­book and oth­ers have in place gen­er­al terms and policies which set out in much detail what shall hap­pen when, and if you want to enter, you must agree to all of them. Even though the applic­a­tion of these rules may seem opaque to out­siders – insiders claim that they observe them with metic­u­lous pre­ci­sion. It does not mat­ter so much wheth­er the rules are trans­par­ent or val­id and enforce­able in phys­ic­al space, in the ter­rit­ory of some state. What mat­ters more is that the for­mu­la­tion, inter­pret­a­tion and applic­a­tion of these rules in cyber­space lie with those, and only with those who impose them.

Rules like this are not pub­lic law; they are the epi­tome of private law in com­bin­a­tion with imbal­anced nego­ti­at­ing power. That private law can play such a dom­in­ant role is any­thing but new, even though it may be a new exper­i­ence for us in our life­time. In his­tory, private law as a reg­u­lat­or, often paired with imbal­anced nego­ti­at­ing power, has always pre­ceded pub­lic law. Because by defin­i­tion, pub­lic law does not exist before there are states and states are always pre­ceded by less con­sol­id­ated, ‘private’ forms of society.

Stock-Check

This is where we are today, with bor­der-cross­ing dis­pute res­ol­u­tion. As far as phys­ic­al space is con­cerned since the late 1940s, when the first inter­na­tion­al courts of mod­ern make appeared; since the late 1950s, when the New York Con­ven­tion came into force as the found­a­tion­al instru­ment for inter­na­tion­al arbit­ra­tion; and since some time later when mutu­al leg­al assist­ance between states became a thing. And since the 1990s when the masses entered cyber­space and brought not only com­merce and R&R, but also disputes.

What is the next bor­der to cross in dis­pute resolution?

We have just crossed one.

As men­tioned, we are part hard­ware liv­ing in the phys­ic­al space that is object­ively on hand, part soft­ware liv­ing in any lala­land that we think up, includ­ing cyber­space (at least we are cap­able to). This is us, humans; some call it Tinker­bell effect. In cyber­space, right behind the bor­der which we have just crossed, we meet creatures that are not both hard­ware and soft­ware, but soft­ware through and through: algorithms.

Here is a cor­rect­ing qual­i­fic­a­tion right away: broadly speak­ing, if we put algorithms into machines, we get robots, which are not soft­ware through and through, but both hard­ware and soft­ware, like us. Very inter­est­ing top­ic, too, how this like us has entered and is con­tinu­ing to enter our tra­di­tion­al phys­ic­al hab­it­at. Whatever can be said about algorithms can often be said in sim­il­ar fash­ion about robots. For the sake of clear-cut reas­on­ing, though, I would like to keep to algorithms and their ‘nat­ur­al’ hab­it­at, cyberspace.

Dispute Resolution with Algorithms

The idea is con­vin­cing: dis­pute res­ol­u­tion using algorithms. I do not mean admin­is­trat­ing dis­pute res­ol­u­tion using inform­a­tion tech­no­logy – this is done since the dawn of IT –, but using algorithms to help resolve. For example by apply­ing the concept of optim­isa­tion, known from sys­tems engin­eer­ing, to set­tle­ment nego­ti­ations and medi­ation, even arbit­ra­tion and lit­ig­a­tion. Or, in arbit­ra­tion or lit­ig­a­tion, by hav­ing algorithms trawl through the massive volumes of case law and doc­u­ments that parties are wont to dis­close these days (‘e‑discovery’). By hav­ing algorithms struc­ture the data and identi­fy pat­terns, trends and lines of argu­ment to pre­dict the prob­ab­il­ity of case out­comes. In oth­er words, by using algorithms to find the needle in hay­stacks of evid­ence that have grown into dens­it­ies and heights that no human can cut through anymore.

(As a side note, this is one way of approach­ing it: find­ing ways to wield big amounts of evid­en­tial data bet­ter. Anoth­er meth­od con­sists of avoid­ing such big amounts of evid­en­tial data in the first place, without com­prom­ising right­ness and justice. But that per­tains to the ques­tion what kind of, and how much, evid­ence is really neces­sary to resolve a dis­pute by arbit­ra­tion or lit­ig­a­tion; a some­what related yet dif­fer­ent sub­ject.)

It would appear that, as of today, algorithms are not often used to optim­ise the res­ults of dis­pute res­ol­u­tion, as a tool for resolv­ers. Rather, today algorithms seem to be used as dis­pute res­ol­u­tion tools by law­yers. Indeed a whole industry has emerged from it, which prom­ises to do to the leg­al industry what FinTech is doing to the fin­ance industry: Leg­al­Tech.

Some say Leg­al­Tech will lessen, per­haps even obvi­ate the need for law­yers doing all these routine works which they have been doing for ages. In fact this has been in pro­gress for some time now. Today, ubi­quit­ous online leg­al data­bases give a single law­yer access to more data at a mouse click than any band of law­yers could have col­lec­ted in the past. Notice­able, how­ever, is the increas­ing pace of algorithms becom­ing smarter. At present, they are still wiel­ded as tools by humans. But what would hap­pen if we should cross that bor­der, too? And pro­trude into an area where algorithms will no longer be savvy tools for law­yers pre­par­ing to appear before a human medi­at­or, or judge, or arbit­rat­or? But where algorithms are the dis­pute resolvers?

Dispute Resolution by Algorithms

All rise, all rise, this court is now in ses­sion! The hon­our­able judge Al Gorithm presiding!

(SCNR. Many apologies.)

This is a bor­der we have not yet reached, let alone crossed. We are not about to either. Wheth­er or when we will cross it seems to be a threshold issue. Dis­pute res­ol­u­tion by algorithms will require of algorithms some­thing which, as of today, they do not seem to mas­ter. Nowadays algorithms can help us find, sort and struc­ture dis­pute-rel­ev­ant data, and fast, and per­haps they can help us pre­dict the prob­ab­il­ity of case out­comes too. Algorithms are tools. To become resolv­ers, algorithms must be cap­able of com­pre­hend­ing both the facts of a dis­pute and the rules applic­able to it com­pletely, and of apply­ing one to the oth­er. This sounds easi­er (I am not say­ing: easy) with a view to lit­ig­a­tion and arbit­ra­tion, at least where the rules applic­able are those of the law. It sounds more dif­fi­cult in view of medi­ation, where the rules applic­able may not be defined so clearly and may com­prise fuzzy things like ‘appro­pri­ate interests’ or ‘con­text’.

In short, algorithms must be much smarter than they are today. Will they ever be smart enough for this? A few rule this out cat­egor­ic­ally while oth­ers think it is pos­sible. Per­son­ally, I believe that in addi­tion to a high degree of arti­fi­cial intel­li­gence, which might be feas­ible, it would require arti­fi­cial con­scious­ness, which was not contrivable.

But who knows? See­ing how cal­cul­able many of us are, object­ively (wheth­er we like it or not) per­haps I’m wrong about the con­scious­ness require­ment. It may prick our pride, but on the whole our needs and wants as well as our man­ners are pretty pre­dict­able and describ­able. Let us assume, there­fore, that an algorithm will not require con­scious­ness to be a dis­pute resolv­er. Let us stay with the smart­ness of algorithms and assume that the bor­der to the smart enough will be crossed someday. Who is going to take algorithms there? Why, their cre­at­ors of course: coders.

Lawrence Lessig has coined the phrase that in cyber­space code is law. He was refer­ring to code as the archi­tec­ture of cyber­space and of its increased import­ance as a reg­u­lat­or, com­pared to the import­ance of law as a reg­u­lat­or in phys­ic­al space. In the same vein he was talk­ing about coders as the (some­times short-sighted) law­makers of cyber­space, but not as the cre­at­ors of algorithms, our cohab­it­ees in cyber­space. Nev­er­the­less, algorithms are the brainchil­dren of coders. Thus, wheth­er an algorithm will be cap­able of com­pre­hend­ing the facts of a dis­pute and the rules applic­able to it, and of apply­ing one to the oth­er, depends at the out­set on its maker.

We have seen that the rules applic­able to the res­ol­u­tion of a dis­pute can be man­i­fold. They can be the law (pub­lic or private), trade prac­tices, social norms, the decision-maker’s own per­cep­tion of fair­ness or a com­bin­a­tion of some or all of these. We have also seen that it may be easi­er to con­vert simple law into code than the com­pre­hens­ive and per­haps inar­tic­u­late interests of the parties. Like­wise, a coder will be hard put to con­vert com­plex law into code, or social norms, or trade cus­toms, or per­cep­tions of fair­ness, at least if the expec­ted res­ult is an algorithm soph­ist­ic­ated enough to be called more than a tool, namely a true dis­pute resolv­er. To reach that level of soph­ist­ic­a­tion, an algorithm might have to eman­cip­ate itself from its maker. In the first place, it has to be able to do that.

Enter self-learn­ing or evol­u­tion­ary algorithms. Sim­pli­fy­ing, these are algorithms cap­able of find­ing solu­tions to a prob­lem by them­selves. Fur­ther­more, they can apply, or learn to apply, a fit­ness func­tion to estab­lish how one solu­tion works com­pared with anoth­er one.

A leg­al dis­pute between, say, two busi­ness part­ners can be a com­plex prob­lem. As of today, no algorithm seems to be able to resolve it all on its own in a sat­is­fact­ory man­ner. A self-learn­ing algorithm approach­ing to do so would have to be able to under­stand the dis­pute and the rules applic­able to it com­pletely, to come up with viable can­did­ate solu­tions and to apply a fit­ness func­tion to each of them. If an algorithm could do that, it would be more than smart. It would have the neces­sary intel­li­gence to decide the dis­pute. Such an algorithm could be a medi­at­or, arbit­rat­or or judge; it could act as the impar­tial third party resolv­ing the dis­pute.

As I said, not a bor­der we have reached yet. But since there are self-learn­ing algorithms, chances are that the day will come when such an algorithm will be intel­li­gent enough to resolve a dis­pute. Just a guess, but this might hap­pen in an area where both the dis­pute and the rules applic­able to its res­ol­u­tion are rel­at­ively simple.

In this con­text, there is anoth­er threshold which should not be under­es­tim­ated. Per­haps it is even high­er than the smart-enough threshold. Dis­pute res­ol­u­tion by algorithms is not only about what algorithms need to be cap­able of. It is also about the parties’ read­i­ness to accept dis­pute res­ol­u­tion by an arti­fi­cial intel­li­gence, without con­scious­ness even. Just a guess again, but this threshold might prove to be lower in cyber­space than in ‘our’ phys­ic­al space, and even lower where the sub­stance of the dis­pute should be cyber, too.

Come to think of it, this accept­ance issue would be avoided, prob­ably, if the dis­pute was not one involving humans but one among algorithms.

Dispute Resolution for Algorithms

Gasp.

The final fron­ti­er of dis­pute res­ol­u­tion. We enter a realm where algorithms are not just tools to resolve dis­putes or even dis­pute resolv­ers them­selves. In this realm, algorithms are parties to dis­putes in their own right.

At the begin­ning, we saw no need to go into detail about what a dis­pute is. We said we know a dis­pute when we see one. But admit it, that was because we assumed a dis­pute was some­thing that occurs among humans. Now that we talk about algorithms being parties to a dis­pute, we can­not ignore this part so non­chal­antly. If we want to think about dis­pute res­ol­u­tion for algorithms, we need to ask: can algorithms be in a dis­pute in the first place? And if so, how do they do it?

Algorithms are self-con­tained step-by-step sets of oper­a­tions to be per­formed to provide a solu­tion to a prob­lem (some say: so are humans). If done right, there is input at the begin­ning and out­put at the end, and often more than one way of reach­ing the lat­ter. If not done right, infin­ite loops may occur. That means a sequence of steps is repeated end­lessly, either because there is no ter­min­at­ing con­di­tion, or one that can­not be met, or one that causes the loop to start over. The algorithm does not come to an end. Last time your device froze – that was one of those loops.

It seems pos­sible to say that an algorithm in a prim­or­di­al stage can have issues, that is: be in a dis­pute with itself (if it loops), but per­haps it is not appro­pri­ate to assume that the algorithm would be able to help itself. To say that such an algorithm could be in a dis­pute with someone, or something else and seek res­ol­u­tion of that would be too far-fetched.

But it may be dif­fer­ent at a later stage of devel­op­ment for an evol­u­tion­ary algorithm, one which can learn and come up with viable solu­tions to prob­lems. It is con­ceiv­able that the solu­tions found by such an algorithm are in con­flict with someone, or some­thing else. In such a case, could we not say that our evol­u­tion­ary algorithm was hav­ing a dis­pute with him, or her, or it?

If we assume that an algorithm, at least one that is suf­fi­ciently intel­li­gent, can end up in a dis­pute indeed, then we can con­tin­ue to ask, as we did before, how such a con­flict could be resolved.

It is inter­est­ing to pon­der what an autonom­ous algorithm might do to resolve a con­flict with itself. Hard to believe that it would go through its issues again and again like its cave­man ancest­ors (algorith­mo­lo­gic­ally speak­ing). It might look for ways and even­tu­ally be able to escape the infin­ite loop, per­haps even with the aid of some­body, or something?

If an algorithm could be in a dis­pute with anoth­er algorithm or with a human being, when would an algorithm give in? On an eer­i­er note, when would an algorithm not give in but try to impose on the oth­er party what it deems to be cor­rect – and how? If it con­sidered itself too weak to pre­vail, would it look for allies or for tools? Would humans qual­i­fy for either and if so, would we be the one or the other?

Would an algorithm seek com­prom­ise and if so, again, how?

Would any­one, or any­thing, be able to inter­vene in a dis­pute between algorithms and act as the impar­tial third party that resolves the dis­pute on its own, by tak­ing away the toy the algorithms are fight­ing about, so to speak?

Would an algorithm be open to inclus­ive dis­pute res­ol­u­tion involving an impar­tial third party, that is to medi­ation, arbit­ra­tion or lit­ig­a­tion? If so, who, or what, would be the medi­at­or, arbit­rat­or or judge? Or would these con­cepts fail, because, for want of states in cyber­space, there is no pub­lic exec­ut­ive and enforce­ment is more effect­ive when done privately? Would algorithms bother?

If any­body, or any­thing, should both­er in the long run, will there be pub­lic courts (or some­thing sim­il­ar) in cyber­space some day, with jur­is­dic­tion over humans and algorithms alike, and some kind of pub­lic exec­ut­ive? How would one enforce any­thing against an algorithm any­way? Regard­ing the enforce­ment against algorithms, Lessig’s code comes to mind again: code as a reg­u­lat­or as sig­ni­fic­ant in cyber­space as law in phys­ic­al space. Pub­lic code as pub­lic authority.

But has not pub­lic author­ity always been part of a state before? Does this mean there will be cyber states some day?

This is dis­put­able, I presume.