Regulating Digital Businesses – Like Chasing Trains

chasing train

I don’t know if you’ve ever had the experience of running for a train that’s just started to move? I’ve had to do it a few times. Yes, I was younger and more foolish then. But it was usually within seconds of the train moving that I was on it. It’s only in old movies that you see the protagonists dashing down the platform as the train picks up speed. Usually, you just have the platform length and the problem is that the train is accelerating. There is a finite window of opportunity after which you’re just going to be left on the platform. This is my very long-winded analogy for regulators and technology. As technology accelerates – it’s getting harder for regulators to keep pace and in fact, in many areas they are just like the proverbial train chasers, running desperately after an accelerating train – often in a futile bid to control a business or industry that is on the verge of leaving the station of regulatory comfort. You can pick from a range of visual metaphors – a man trying to control seven unruly horses, or grabbing a tiger by the tail, but you get the idea. Regulators are in a fix.

The sight (and sounds) of the congressional hearing of Mark Zuckerberg did not bode well for regulators. They should have had Zuckerberg dead to rights over it’s (willing or unwilling) culpability in the Cambridge Analytica imbroglio. Yet he came out with barely a scar to show for 2 days of grilling. Many of the people asking him questions came across as the stereotypical grandparent trying to figure out the internet from their grandchild, even if these are very exaggerated caricatures. There is arguably a 40 year age gap between the average lawmaker and the average entrepreneur. But the age challenge is just a minor problem. Here are some bigger ones.

Technology businesses are shape-shifting enterprises invariably redefining industries. Platforms cannot be regulated like their industrial counterparts. Uber is not a taxi company. Facebook is not a media business. Airbnb is not a hotel. No matter how convenient it might be to classify and govern, or how often someone points out that the world’s biggest taxi company doesn’t have taxis. No, these are data and services platforms, and they need an entirely new definition. You could argue that the trouble with Facebook has come about because they were being treated like a media organisation, rather than a data platform. And let’s not forget that the only reason Facebook was in the dock is because of the success of Cambridge Analytica in actually influencing an election. Not for the misuse of customer data on a daily basis which may have gone on for months and years by Cambridge Analytica as well as other similar firms. While governments’ focus on Uber stems largely from incumbent and licensed taxi services, nobody seems to be worried that Uber knows the names, credit card details and the home and office residences of a majority of its users.

Tech businesses, even startups, are globally amorphous from a very early age. Even a 20 person startup barely out of its garage can be founded in California, have it’s key customers in Britain, its servers in Russia, its developers in Estonia and pay taxes in Ireland. Laws and governments are intrinsically country bound and struggle to keep up with this spread of jurisdiction. Just think of the number of torrent services that have survived by being beyond the reach of regulation.

These are known problems and have existed for a while. Here’s the next challenge which is a more fundamental and even an existential one for lawmakers. With the emergence of machine learning and AI, the speed of technology change is increasing. Metaphorically speaking, the train is about to leave the station. If regulators struggle with the speed and agility of technology companies today, imagine their challenge in dealing with the fast-evolving and non-determinate outcomes engendered by AI! And as technology accelerates, so do business models, and this impacts people, taxes, assets, and infrastructure. Imagine that a gig-economy firm that delivers food home builds an AI engine that routes its drivers and finds a routing mechanism that is faster but established as being riskier for the driver. Is there a framework under which this company would make this decision? How transparent would it need to be about the guidance it provides to its algorithms?

I read somewhere this wonderful and pithy expression for the challenge of regulation. A law is made only when it’s being broken. You make a law to officially outlaw a specific act or behaviour. Therefore the law can only follow the behaviour. Moreover, for most countries with a democratic process, a new law involves initial discussion with the public and with experts, crafting of terms, due debate across a number of forums and ultimately a voting process. This means we’re talking in months, not days and weeks. And if technology is to be effectively regulated and governed, a key challenge to address is the speed of law-making. Is it possible to create an ‘agile’ regulatory process? How much of the delay in regulation is because the key people are also involved with hundreds of other discussions. Would lawmaking work if a small group of people was tasked to focus on just one area and be empowered to move the process faster in an ‘agile’ manner? We are not talking about bypassing democratic processes, just moving through the steps as quickly as possible. A number of options are outlined in this piece from Nesta website – including anticipatory regulation (in direct contravention of the starting point of this paragraph), or iterative rather than definitive regulation. All of these have unintended consequences so we need to tread cautiously. But as with most businesses, continuing as present is not an option.

Then there’s the data challenge. The big technology platforms have endless access to data which allows them to analyse them and make smarter decisions. Why isn’t the same true of regulators and governments? What would true data-driven regulation look like? We currently have a commitment to evidence-driven policymaking in the UK (which has sometimes been unkindly called policy driven evidence making!) but it involves a manual hunt for supporting or contradicting data, which is again time-consuming. What if a government could analyse data at the speed of Facebook, and then present that to the experts, the public, and legislators in a transparent manner? The airline industry shares all the data about every incident, accident and near miss, across its ecosystem, competitors, and regulators, and this is a significant contributor to overall airline safety. (Outlined in the book Black Box Thinking, by Matthew Syed.) Why isn’t the same true for cybersecurity? Why isn’t there a common repository for all the significant cyber attacks, which can be accessed by regulators armed with data science tools and skills, so that they can spot trends, model the impact of initiatives and move faster to counter cyber attacks? If attacks seem to originate from a specific territory or impact a specific vulnerability of a product, pressure can be brought to bear on the relevant authorities to address those.

running after train
These are non-trivial challenges and we need to be aware of risks and unintended consequences. But there is no doubt that the time has come for us to think of regulation that can keep pace with the accelerating pace of change, or governments and regulators will start to feel like the protagonists of movies where people run after trains.


Resisting Technology Is Like Resisting Ageing

My wife (Karuna) and I often differing views on a number of things, as is common. And almost always, she’s right. But there are some areas where we agree to disagree. 
Karuna doesn’t drive a manual car. She’s very comfortable in an automatic. I love driving – either manual or automatic. Obviously, the automatic car is doing a whole lot of thinking for you. And probably doing a few things better. By matching the gear to the speed more effectively, it’s likely to be more fuel efficient especially in stop-start city driving. But like most people who drive a manual car, I hunker for the control of the stick shift and the level influence I have on the drive. It feels like I’m closer to the engine. The automatic car provides a level of abstraction and let’s anybody drive, without mastering the intricacies of gear shifts and clutch control. Somewhere in the recesses of my mind, the message keeps flashing: the automatic is all right, but a manual car is a real drive. 
We have the opposite stances on Digital Cameras. As somebody who has formally learnt photography and spent time in dark-rooms developing prints, she loves the control, and human input into the process. I enjoy the fact that I can get great photographs by just framing the picture. Karuna gave me tips on framing but the camera does the rest – i.e. manage exposure, focus, lighting, and even the intensity and balance of colours. Of course, all of this comes bundled with a phone. No more wandering around with an SLR camera slung around your neck. I love it. For her its anathema. 
The pattern here is simple, when we invest time and effort in building a skill, or a technique, we are invested in the process, not just the output. And what almost every technological advancement tends to do, is that it democratises is previously closely held skill, putting the same level of competence into the hands of amateurs and novices. For the experts this is distasteful or downright annoying, but more importantly, it’s often professionally disruptive. The former, because it devalues that expert process which we are attached to, and the latter, because it challenges their expertise and renders them less valuable. 
“The Knowledge” is the course that all London Black Cab drivers go through. For decades, the London Cab has been famous – one of the icons of the city. Apart from the car itself, which is custom designed and manufactured for the purpose, the drivers are famed for their familiarity with the city and routes. The Knowledge comprises some 320 routes through London, and covers 25,000 streets and 20,000 landmarks. A black cab driver is expected to know them all. Qualifying takes 2-4 years on average. During the exam, they can be given any start point and end point in those hundreds of routes and they are expected to know the most efficient way of getting from start to finish. The number of qualified drivers is controlled. Typically, it takes an investment of 30,000 to become a cab driver, in addition to the 25 hours a week time invested over 3 years.  Typically, the London Cab is twice the price or more for journeys that take 30 minutes or more, compared to the privately run ‘mini-cabs’ that also operate in an organised manner in London. 
Since the dawn of sat-navs any driver can find locations, routes, and optimise journeys with an investment of under a hundred pounds. Nowadays the smartphone does just as well. Today every user who gets into a taxi is more likely than not to have a device with him or her that can provide exactly the same level of knowledge about routes, directions and traffic conditions that the black cab driver has accumulated over 3 years. Short of injecting this knowledge into the brain, a la Matrix, the first time tourist in London is now as well equipped to navigate London as the black cab driver. 
Of course, you still need to get a taxi, and the black cabs are ubiquitous in London so you’re likely to hail one anyway. Or you would, till the arrival of the brigade of taxi apps. And the poster child of taxi applications – Uber. Now you just send up a digital flare while you’re working your way through dessert and you can be sure that by the time you’re out on the street, the taxi is likely to be there. Not a London Cab but a less expensive car with a similar assurance of safety and comfort. 
Not that London Cabs are luddites. The Hailo and Gettaxi  pps do exactly this for black cabs. The whole experience of calling a taxi has changed forever. You just broadcast a request and one of the many taxis which is the closest to your location responds. It’s the same for any category of cabs. Even the cab companies which take bookings do so through apps. It’s just that the price premium charged by London cabs is no longer sustainable. 
There are plenty of other services run by local cab companies which come with Apps. I use a company called Swift  which has a reliable app and also one of the drivers, let’s call him Bob, asks me about it whenever he picks me up. The last time around we had a discussion about some of the features that the app should add. He is very engaged with the idea of the app making this experience better. 
As I write this, all over the world, incumbent taxi services are warring with new services such as Uber and Lyft. Which are by the way, just marketplaces, and not car services, themselves. And clearly much of the legislation does not cover this model. So the incumbent services are lobbying the government for protection. In Germany, a cab license costs over $ 250,000. Understandably, drivers having paid that sum are not happy to see their returns diminished via competition from new and technologically enabled entrants. Many cities including Munich, Dusseldorf, Berlin and Hamburg are considering declaring Uber illegal. Their argument is primarily that as taxi services, Uber enabled cabs should pay the same license fee. 
In Seoul, the government’s concerns are based around the safety of the vehicles, background checks on drivers, and the impact on the local taxi trade. The last may be the most honest reason, in most parts of the world. Even though in Seoul, Uber is more expensive than the regular taxis. 
Even at home, in the US, Uber has faced the law – in Virginia for example, where Uber has been asked to ‘cease and desist’ by the government, till it obtains the ‘proper authority’. 
Brussels has already banned Uber. Barcelona, Paris and other major European cities have discussed banning it. There have been strikes in London and Milan. All of these are typically examples of old markets and legislation trying to keep up with new business models. Even Neely Kroes has criticised the bans.
The pattern that repeats itself is that markets switch quickly, but legislation takes time. Most taxi apps now allow sharing, payments, and a host of other features which significantly improve the experience for the user. 
Defending the old model even in the face of new technology creates a precipice from which the fall can be sudden and dramatic – witness the music industry, which reaped the benefits of digital technology for many years but failed to adapt to the internet’s new models. People find ingenious methods for using the new technology to the benefit of suppliers and customers, even as regulators and enforcers fume. 
So where does that leave the Black Cab driver who has just spent years mastering “The Knowledge” to qualify to drive a black cab in London? Is this the end of the road for him? Is this one more example of technology rendering a valuable skill useless? 
Your guess is as good as mine, but for a glimpse of what could happen, let me take you back a hundred and fifty years or so. It was the time of the invention and spread of photography. I’ve written about this in more detail here but the short version is this: photography democratised portraiture. And rendered hundreds of artists jobless. Any amateur armed with a camera could take a photo more accurate and lifelike than the best of painters. So what did these artists do? Many presumably changed professions, some undoubtedly fell on hard times. But out of this some decided that their role was not to represent reality but to interpret it. It is no surprise therefore that the birth of impressionism coincided with the spread of photography. 
So when democratisation hits your area of expertise, as it will, sooner or later, will you find yourself with a choice of extinction or adaptation. Will you be like the impressionists and evolve? Or will you fall on your sword (or paintbrush)? Will you look for help to regulators? Or will you create new markets? After all, even decision making and ‘management’ expertise, is being democratised through analytics and knowledge systems. 
Either way, the challenge for regulators as always, is to move at the pace of technology and markets. The challenge for you is to evolve to find or create a market as technology democratises your specialist skill. Resisting the change, though, is not really an option. You might as well try to resist ageing.