Thought Leadership

Image
Dr. Horst Walther
Let algorithms rule – not politicians!

By Dr. Horst Walther, DC Commissioner for UN Affairs, Principal The Diplomatic Council Otto Schell Institute for Digital Transformation (DCOS)

There are many complaints to be heard about the behavior of politicians, the low quality of their decisions, their selfishness and their lack of vision. Some elected leaders of world powers are even accused to be morally unfit for their job.

Although – not only – according to a recent cover story of the time magazine, the “strongmen” are on the rise worldwide, their unpredictable, largely incoherent emissions meanwhile are widely feared – once they are elected into their respective offices. Populist ad hoc decisions lead to discrimination of minorities up to outright genocides, regional wars or the implementation of “voodoo economics”, or they are simply ruining the national budget entrusted to them in an helpless attempt to fulfil their insane campaign pledges. In short they usually do more harm than good.

Although these complaints seem to be as old as our civilisation, let us assume for a moment that the allegations are somewhat well-founded.

1. Shifting beliefs and forces

There are disturbing indications to be observed, which have the potential of shattering widespread deeply entrenched beliefs like those that (1st) human beings have the capacity to govern themselves and that (2nd) we hold the inherent dignity to deserve self-government.

They find their strong contradiction in the undeniable fact that around the world strongmen are seeking unchecked power and find mounting success in gaining it. Polling in the United States and other developed nations suggests increasing openness to the idea of authoritarian government, especially among younger people. (According to the World Values Survey, almost one-fourth of U.S. citizens ages 16 to 24 said that a democratic system was a „bad” way to run the country in 2011, about twice the percentage as among those over 65.)

The last straw that breaks the camel’s back may be seen in a the condensed view of Nathan Schneider, a reporter and professor of media studies at the University of Colorado, Boulder voice in a preview of his new book “Everything for Everyone: The Radical Tradition that Is Shaping the Next Economy”:

“The assumption that liberal democracy is the eventual destination of historical progress can no longer be taken for granted.”

Besides that he sublimely conveys a questioning of the whole notion of an ongoing and unidirectional human progress (which is a topic for later elaborations) he makes a clear point on the necessity of leaders:

“We the people are more connected than ever before, more capable of managing our own economies and collaborating across the globe, and yet we stand at the mercy of petty men with nuclear weapons, flinging insults across oceans. We are better than this. Perhaps we are finally ready for a politics in which the president matters less.”

This remark makes an initial dent to our mainstream beliefs that nation have to be led by leaders hereby paving the way for more radical views to follow. But first let’s take a look at the players in the field and the game itself: The leaders and their followers, and the process of decision making.

2. Career politicians – The human factor

What do politicians do? Politics would be the obvious and simple answer. But is that really the case? Or are they just plainly looking for a seat at the table of the mighty and the powerful? Pursuing political goals or simply making career?

Typically there are two main career paths which are typically followed by politicians seeking careers in modern democracies.

First come the “career-politicians”. They are politicians who work in the political sector until retirement. Doing politics is a job for them. Being very ambitious they intend to excel in their job and trump all the others – quite often regardless of the message, the political content they are to bring forward.

“The biggest threat to politicians is when everything turns out to work fine without us being involved. Don’t let that get around.” Sigmar Gabriel, then Federal Minister for Foreign Affairs, stated in his speech at the 2018 New Year’s Reception.

Even if he explicitly wanted it to be understood as ironic, it stuck to my memory. It perfectly sums up the plight of a career politician: if his name cannot be permanently associated with effective and positive actions in his area of responsibility, his image will fade in a population saturated with information.

He is not allowed to relax on seeing his political goals achieved and being bored then contentedly retreat to his estate. Just as Otto von Bismarck, who is quoted as “I am bored. The great things are done. The German Reich is made.”

He has to prove his raison d’être every day. Because that’s his job, for some of them, the only job they have ever acquired skills in.

Second are the “political careerists”. These are politicians who gain reputation for expertise in controlling certain bureaucracies or even more prominent functions, then leave politics for a well-paid career in the private sector making use of their political contacts.

One of the more spectacular examples to date of a Western politician selling out to an even authoritarian government may be seen in the nomination of former German Chancellor Gerhard Schröder to the board of Rosneft, Russia’s largest oil company.

In both cases and for obvious reasons “career” is used as a derogatory term. The implication is that the person is only interested in obtaining positions of political power. They spend all of their efforts in appealing to the masses and will do/say/promise anything to win support and always try to be part of the winning party.

3. Charismatic leaders – two edged swords

Are there no true charismatic transformative leaders? Oh yes there have been quite a few throughout history. Archie Brown discusses some of them in his landmark book “The Myth of the Strong Leader: Political Leadership in the Modern Age”.

Although politicians in the sense of men or women obsessed by a certain political goal, they regard as mandatory to pursue by whatever means it might take, are more often found at the right or left edge of the political spectrum, there are some illustrative examples to be found in the political mainstream as well.

Among those choosing the means, perhaps Winston Churchill took to the extreme by switching political parties (“ratting” in Parliamentary terms), which he did not once but twice. He left the Conservatives (Tories) for the Liberals in 1904, only to re-join the Conservatives in 1925 for which he has long been criticized for.

Churchill defended his actions: “The only way a man can remain consistent amid changing circumstances is to change with them while preserving the same dominating purpose.” So his argument was that only by switching alliances he was able to achieve his goals.

These arguments point less towards career making purposes but rather to pursuing political goals. One is tempted to cite Otto von Bismarck to his defense, defining politics as the the art of the possible[1]. Only on this winded course his political goal was achievable, was possible.

But nevertheless unease might arise as we could recognise too much dependence of the political business on individual personal decisions in this case – as in many others.

This uneasy feeling is not limited to myself and it is not completely new. Already back in 2012 Tomas Chamorro-Premuzic wrote in the Harvard Business Review on “The dark side of Charisma”.

Charisma fosters collective narcissism

Charisma is addictive

Charisma disguises psychopaths

Charisma fosters collective narcissism

So not only that we may be unlucky enough to let our political system churn out career politicians, who are unwilling and unable to cope with upcoming challenges. We may on the other hand fall victim to strong, charismatic transformative leaders.

Why do we become victims so easily? Well, good question and to be discussed next.

4. The followers – those who are led

There are those who lead – the leaders – and those who obediently allow them to be led – the followers. Although in liberal societies by definition the people ought to be the true sovereign and the politicians their agents there seem to be an underlying archaic trait deeply rooted in mankind’s soul: the longing for a strong leader.

This observation, the well-known the myth of a strong leader, seems to hold not only in young democracies where one can assume an authoritarian leadership style being deeply imbued into the fabric of the society out of mere tradition.

„We need a strong leader to lead us out of the political morass.“ During my years in Ukraine I heard this sentence quite often. “If you want a strong leader, you will get a strong leader – along with all the ugly calamities, he will cause,” was my usual response. Strong leaders usually become such by relentlessly pushing their very personal agenda forward, leaving behind a trail of devastation and scorched earth.

Still not really thinking in terms of an egalitarian democracy and rather used to a culture of a high ‘power distance’ and authoritarian leadership style, my dialog partners revealed that they in fact were longing for some kind of benevolent dictator. This has rarely worked as expected.

But even western democracies are not immune to this fatal longing. We should better be watchful not to slide down that slippery slope towards the road to serfdom again and again.

5. Time for more direct democracy?

So, let’s risk more democracy? The electronic communication channels of this hyper-connected world provide new means for opinion creation and political decision making – so the theory goes. Remember that Estonia, that small nation at the shores of the Baltic Sea offers participation in national elections via an e-voting channel.

Switzerland on the other hand is known for its semi-direct implementation of democracy – imagined halfway between a representative and a direct democracy. Next to the election of the Federal Assembly it offers mandatory referendums, optional referendums and federal popular initiatives.

Why not combine both worlds and provide a way of direct participation of every citizen in the daily political business – very much this is done in today’s parliaments in the more representative style? We the people are the ruling sovereign? Wouldn’t this represent the perfect form of democracy?

Sounds good – is dangerous however.

Unfortunately, due to the very nature of mankind it is not as simple as it may appear. The easiest way is the original way of voting, the vote in a hall, under a baobab tree or in the Thing.

Here, every participant, originally every free and honest man of full age and good standing, had a vote with which he (or occasionally she) voted for or against a specific proposal at a given time.

Things can go wrong here too. An inciting, inflammatory speech, given immediately before voting may have achieved its demagogic purpose and led the voter to make a hasty decision, which he might have preferred not to make after some reflection.

On the other hand however, one can safely state that democracy in this original form may work reasonably well: Opinions are expressed, all participants have equal access to information during the time, are present in the same way, have equal chances, will abstain from voting only for good reason. And the votes can be counted in front of everyone’s eyes if necessary. In case of secret ballots, we can even assume free and independent decisions.

As the group grows, the choice becomes more difficult to organize. Also the governmental work that may be performed becomes more complex and difficult to understand. That is why the various forms of representative democracy have developed. Here we delegate trustworthy people from among us to whom we entrust this complex work. They represent us and decide on behalf of us. This representative democracy has obvious advantages and appeared to be the only option for larger communities, at least until now.

However, this form of governance also has its likewise obvious, serious disadvantages – as discussed above. In addition its actors may become more or less easy prey to lobbyists of a particular interest.

Isn’t there a better way? Don’t we today have the technical means to hold (electronic) elections in larger groups? So it’s time for more democracy?

That’s a nice idea. Unfortunately, it rarely works.

Referendums, whether held traditionally by electronic means, can completely get out of hand and run counter to the original idea of more direct democracy. After repeated votes quickly, a certain electoral fatigue sets in. The voters called to vote stay away from the ballot boxes. All too easily, non-representative but resolutely acting groups can take advantage of this circumstance for their purposes.

So, direct democracy is a sharp weapon. It requires a high sense of responsibility and a high degree of social maturity. Doubts are justifies that many peoples already have arrived at that level of this maturity.

Disillusionment has not only arisen with regard to direct democracy. The electronic communication channel in its current form has also fallen into discredit and has even lost the support of its early prophets.

To emphasize this point, e.g. the well-known VR pioneer Jaron Lanier mentioned in a conversation on Silicon Valley’s politics and what went wrong with the internet: “One Has This Feeling of Having Contributed to Something That’s Gone Very Wrong” and “That it’s ruining politics, it’s empowering the most obnoxious people to be most influential, and that’s destroying the world.”

It looks like we are caught between a rock and a hard place here – truly a dilemma.

6. The odds of decision making

Not only are decision-makers not as reliable and trustworthy as necessary, whether due to hidden agendas or one of the multiple biases, or as in the case of decisions made by the “crowd” because they are distracted, emotional, lack information or simply bored. Decision making in itself seems to take us humans to our mental limits.

MIT’s McAfee Andrew stated in an interview with McKinsey & Company in March this year: “A lot of executives who I talk to think that a big part of their job is making the tough calls; relying on the experience, industry knowledge, and judgment that they’ve built up; and having a clearer crystal ball of things that are going to happen in the industry or in the future than other competitors have.

While I value those things, I value them a lot less than I used to, because … over again we’re seeing that technology is better at human-judgment tasks than humans.”

Next quote is taken from the Farnam Street Blog on Decision Making: Do Algorithms Beat Us at Complex Decision Making? where the author states “In the battle of man vs algorithm, unfortunately, man often loses. The promise of Artificial Intelligence is just that. So if we’re going to be smart humans, we must learn to be humble in situations where our intuitive judgment simply is not as good as a set of simple rules.”

The author in turn makes extensive use of the groundbreaking works of Nobel laureate Daniel Kahnemann, who indeed is the central expert whose work to draw from when it comes to decision making – might it be human or algorithm based. In is landmark publication “Thinking Fast and Slow” he elaborates on the various biases humans fall victim to when trying to make decisions in a complex problem space. I too made use of his insights before – applied to the realm of corporate decision making, where much research has been done.

Human irrationality is Kahneman’s great theme. A leitmotif of this book besides all the cognitive biases, fallacies and illusions that he and Tversky have discovered in the last few decades, is overconfidence, nurturing the impression that humans are fundamentally irrational.

This impression is not naturally true, like Kahnemann himself makes clear. We just have to take into account out built-in constraints we face when confronted with decision needs of a complex nature.

Under such conditions, human decisions, even if they are unbiased and well-intentioned, suffer from considerable random errors, which Kahnemann aptly calls noise. And noise was exactly what he found while consulting large corporations: “I’ve been observing more folly than I expected,” he says. “People often say that the private sector is better than government, and if government is worse than what I’ve seen, then we’re really in trouble.”

Yes, probably we are.

7. So better let algorithms decide?

Okay then, let’s make it less human-bound, less susceptible to human peculiarities, weaknesses, and noise. No longer dependent on the political genius, the super negotiator, the strong and charismatic leader.

As we have learned, we are easily deceived. We are often biased, hereby bending toward a more personal benefit in terms of personal honors and even outright financial benefits for ourselves or our buddies. And to make matters even worse, we quickly reach our limits. We make noisy decisions – simply because they can’t do better.

Well, for now, people (still) have to be responsible for building the rule set, formulating the algorithms, defining the driving policies and determining the underlying values. But drawing the right conclusions should be the domain of clever machines. Will it improve our current situation? Or will it prove dangerous?

So what are the perils we have to face?

Not drawing the right conclusions from our rule set? No, I have no doubt and so haven’t others, that a good decision engine will be able to evaluate a given set of rules / algorithms of whatever complexity as long as they are not contradictory and are fed with sufficient information of an appropriate quality.

The two minor and innocently looking side conditions consistency and completeness pose tough challenges already. However they do so for human decision makers as well. Humans will have to rely more on their intuition, the decision making system #2, how Daniel Kahnemann would call it. As he had delivered proof, humans are not better at this than machines.

Not having the right rules at hand? Well this is much more of a problem. Finding general rules means abstraction, generalization, induction or deduction, tasks which proved to be difficult for us.

So, we have to face new challenges, right?

8. Rule based decision making – a new idea at all?

So is this a groundbreaking, revolutionary new idea? Not at all – to start with the conclusion right at the beginning. It is so evident that there can be no doubt. It probably all started with the codification of law. The Code of Hammurabi for example dates back to 1790 BC and it might even not be the earliest attempt to replace arbitrary court decisions by the rule of law – just among the best documented and most comprehensive ones. It is pre-faced by the poetic words “Anu and Bel called by name me, Hammurabi, the exalted prince, who feared God, to bring about the rule of righteousness in the land, to destroy the wicked and the evil-doers; …”.

Quite a bit later Aristotle is quoted with “It is more proper that law should govern than any one of the citizens”. The “modern” version of a civil code, the Corpus Iuris Civilis (“Body of Civil Law”), emerged already 529 to 534 under Eastern Roman Emperor Justinian I. The concept of codification took a further boost during the 17th and 18th centuries AD, as an expression of both natural law and the ideas of the Enlightenment. It was the common belief at this time that all spheres of life could be dealt with in a conclusive system based on human rationality, following from the experience of the early codifications of Roman law during the Roman Empire. The so called “common law countries” of the United Kingdom and Ireland and Scandinavia however remained untouched by the codification movement.

So – to a varying degree – there is a tradition to follow rules. While both – rule creation and rule application – were performed by humans, hereby allowing for a large margin of discretion – they nevertheless represent the common ground on which the community agreed as a framework for future decisions.

9. Transparency – an unwanted side effect?

It also turns out that we tend not to be honest about what really drives our decision-making. And that’s not just an option, it’s a fact. In his book “The Tragedy of Great Power Politics”, published in 2001, the American scientist John Mearsheimer convincingly shows that double standards were regularly applied during the Cold War and probably beyond. While the US often rightly pointed to human rights violations while simply pursuing power politics, the Soviets were reluctant to be blunt in their actions as well and called their actions a promotion of communism. However, both sides could easily have been caught at the forefront of human rights violations or violations of their communist principles.

If algorithm based decision making will be able to earn any trust, it needs to be radically transparent. Maybe future practice will teach us otherwise. However my current expectation is that such translucent process will make decision making brutally honest. May be this will cause some collective learning effects. So double standards will either have to become publicly visible and hereby accepted parts of the rule set or cannot be part of the political tool set any longer. While I obviously tend to the latter, I am not so sure if public opinion could not even come to terms with the former.

As I put hope over fear, I want to express the expectation that an algorithmic decision making will lead to more transparency and thus to the avoidance of double standards.

10 Risks & Fixes

The vanguard it security pioneer and inconvenient warner Bruce Schneier already in 2016 addressed the topic of “The risks – and benefits – of letting algorithms judge us”. He pointed at the then still early attempts of social profiling by the Chinese state, resulting in a universal well-behaving score. That’s the Chinese way of doing things – and they even have a long tradition in these attempts – and now even the appropriate technology at hand.

The free liberal American way of doing things however isn’t less daunting. Although here the scoring activities – as it is common custom – are left to privately held corporations. Opt-in mechanisms and the freedom of contracting nourish the illusion of a voluntary participation. However in practice the choice would be to drop out of the society – or to comply with the opaque rules. And of course there can be now doubt that stately actors will gleefully welcome the opportunity to grab the profiling info for their own, not substantially more, transparent purposes.

His conclusion was “The first step is to make these algorithms public.” Of course transparency is such an obvious requirement and precondition for a just governance of a liberal democracy that we hesitate to name it in the first place. We tend to take it for granted for all activities performed on behalf of the true sovereign – the people itself. However this is far from being reality and not even in the genuine interest of those who switched sides from those being ruled towards the ruling ‘class’.

Bruce Schneier rightly adds a second advice “The second step is for these systems to be subject to oversight and accountability.” Innocently embedded in this claim is another very useful link to “Algorithmic Accountability: On the Investigation of Black Boxes”. In short, next to being public these rules need to be approved, scrutinized and re-examined regularly. As they express how we intend to rule ourselves, the governing of these algorithms central to the democratic decision-making process

11. Acceptance

Resistance is to be expected, in particular by those who are going to be replaced by those smarter machines, as Michael Schrage observes: “In theory, the organizational challenges of algorithmic autonomy map perfectly to which processes or systems are being made autonomous. In reality, transitions prove to be significant operational problems … creates interpersonal and inter-process frictions. At one American retailer, an autonomous ensemble of algorithms replaced the entire merchandising department. Top management told store managers and staff to honor requests and obey directives from their new ‘colleagues’; the resentment and resistance were palpable. Audit software and human monitors were soon installed to assure compliance.”

If not personally affected negatively however even among the Germans, the well-known laggards in terms of leading edge technology adoption, some acceptance for algorithmic decision making seems to be growing. At least the result of a survey among 1,006 German citizens aged 14+ conducted by the digital association Bitkom supports this view.

According to that survey a majority of the German citizens would even let an ‘AI’ decide on their behalf in certain situations. Not sure if all participants had a good understanding of what’s AI and how it differs from traditional rule evaluation engines.

Regardless of these subtleties, about 15 percent say they would rather prefer an AI’s decision over that of a human while applying for a loan at the bank. 10 percent would accept a non-human judge in court, e.g. after a traffic accident. Another 9 percent would rather negotiate their next salary increase with an AI than his / her boss. This said, in total 6 out of 10 German citizens (58 percent) would prefer an AI decision at least in certain situations.

I therefore expect the greater resistance from the governing side, which would have to stop using double standards, backroom negotiations, secret agreements and other usual dark practices.

12. Conclusion

In contrast to the omnipresent politician approach as satirized by the Sigmar Gabriel quote above, the algorithmic approach aims exactly at the opposite: doing the political business as automated, silent and in the background as possible. It ought to be fully transparent and logically traceable. Abstaining from exciting spectacular show effects, it should rather appear as a dull routine. In all cases when there are no crises to overcome, governance should be boring. That would even be a quality criterion.

The real work of a politician would then consist of reviewing these decisions, refining or streamlining the rules, or at least adapting them if necessary – but not participating personally in the decisions. That would be considered as archaic by then. And certainly not should he personally appear as leader of the nation, along with first Lady, first Daughter and first Dog.

Jack Ma, founder of Alibaba Group, China’s e-commerce giant mentioned on World Economic Forum Annual Meeting 2018 in Davos that “People will never be able to compete with machines”. So it is just natural that AI will take over our jobs. Certainly this prophecy applies to the more mundane tasks of a politician’s daily life too.

However, this may only be a transitional state until an Artificial Intelligence (AI) completely takes over all government works. By then all politicians may safely retire completely. Because in this vision everyone will be a politician then again.

This would be or – to be precise: will be – truly a digitally transformed governance. It’s not just some far-off dream / fear anymore. It is not about indulging black magic and alchemy but rather quite the opposite: forcing those in charge of governing us on behalf of us to follow a predefined rigorous logic. The rules need to be set by us humans like it as a long tradition since the times of Hammurabi. The conclusions however should better be drawn by a digital processor rather than a human one.

Although I personally consider it a little premature to elect a robot as a city mayor, who already exposes a modest degree of artificial intelligence (ai), which however can safely still be described as artificial stupidity (as).

Nevertheless I can follow another Nobel laureate Richard Thaler, who is quoted “We’re just scratching the surface on what technology can do.” in the context of “Making better decisions through technology”.

Digital transformation of governance seems to be on its way already.

[1] „Politics is the art of the possible, the attainable – the art of the next best“, – Otto von Bismarck