This summer, we have visited Yellowstone. I couldn’t resist the temptation to do so in a convertible Mustang, which was the exact right choice. Coming from Salt Lake City, we left the highway at Ogden to climb an alpine, narrow and winding road which led us on a barren, hilly plain — where we were glad that ‘convertible’ also means that you can close the roof, as temperatures had dropped considerably — to head for the Southern entrance of Yellowstone, passing the impressive peaks of Teton Range. In Moran, however, we turned East to follow the Wind River valley to then, via Dubois and Cody, enter Yellowstone through the Western entrance at Wapiti. Via Booking.com we had organized our trip into convenient chunks of driving which, up to then, involved stays in Ogden, Thayne, and Dubois. The American institution of the motel is, of course, a result of this kind of motorized travel and I really like it. It is convenient, understandable and, to me, a reflection of the American society and culture: individual units arrayed into neighbourhoods which cater to both one’s desire for withdrawal from society and one’s need to engage with each other. However, in Wapiti something unexpected happened.

As we tried to find the aptly named Cutthroat Guest Ranch, we were first struck by a sign prominently attached to one of the poles of the typical three-beam farm gate, which announced that this property was up for sale. We then followed a steep gravel road, not at all suited to our smug car, to approach a seemingly deserted place. The log huts, arrayed in that characteristic manner, i.e. standing at a friendly distance from one another, were not inhabited and neither people nor vehicles could be seen apart from a dusty truck parked a bit off-site. The reception room was locked and no sign of activity could be spotted. We tried to call the owner — there was a telephone number given in our reservation confirmation — but could only reach an answering machine. Since our booking confirmation had warned us that check-in is possible only after 4 p.m., we had timed our arrival accordingly, so that we became quite worried as to where to stay for the night. Luckily, we got hold of a cellular signal to make connection with the virtual world again to find a nearby motel, a little closer to the Western entrance of Yellowstone, located in a pretty valley with a reception hall decorated like a living room and inhabited by large friendly dogs and TV screens that beamed old cowboy movies at guests round the clock. There, I could also check my email and, to my surprise, found instructions by the owner of the Cutthroat Guest Ranch how to enter our cabin, explaining why there was no personnel on site (because their main business now consisted of running a restaurant in Cody). What ensued was a polite but unfriendly exchange in which I argued why we had been right to assume that their place had been abandoned while the owner argued that the procedure of remote check-in, announced on the day of arrival through email, was completely aligned with standard practice, shaking my confidence in the American institution of the motel. Back in Germany – after an enjoyable exploration of Yellowstone from Gardiner and Bozeman and a return to Salt Lake City along its Western flank via Tetonia – I then found, not entirely unexpected given our prior email conversation, that my credit card had been charged for the (unused) night at Cutthroat Guest Ranch. I now turned to Booking.com for advice.

The first thing I learned was that there is no easy way to contact them. Finding a telephone number was a challenge and then my patience was tested as I counted down my position in the telephone waiting line. In total, I talked to them three times, each time starting all over to tell my story and to win over my counterpart by explaining the situation in which we had found ourselves when we arrived at the Cutthroat Guest Ranch. The second person I talked to, a woman with an East European accent, sympathized with me and promised to take tough action. However, my last human encounter with Booking.com, with a man also with an East European accent, settled the matter. He explained to me that Booking.com has, indeed, contacted the owner of the Cutthroat Guest Ranch to suggest that he refunds the rate for the unused night – to no avail, of course – but that they could do no more than appeal to his honesty and sense of good business manners. My expectation was that Booking.com would take responsibility for the (non-) service offered through its platform, but I learned that this is not the case.

They standardize the process of how the various parties interact with each other and their responsibility ends when that process has been complied with by all parties. In my case, this would have involved that I had contacted Booking.com when we arrived at the locked and seemingly abandoned guest house; of course, the process defined by Booking.com does not involve the definition of a maximum waiting time in the telephone line (I had spent 45 minutes in that waiting line when I first called them). Booking.com would then have tried to contact the owner and given him 40 minutes to resolve the issue. If by then no solution could have been found – not an unlikely outcome given that we had only reached an answering machine when trying to contact the owner ourselves – they would have declared the reservation cancelled, so that by then we would have been free to find another place. This would have been close to 6 p.m., having involved anxious waiting in a deserted place of up to two hours.

Now, you might say that this is not a particularly bad experience and every traveller has to be prepared to encounter this kind of situation. And you would be right. I wrote off this experience by categorizing the event as such, things that are to be expected on self-guided exploration tours in an unfamiliar environment. Actually, that is an important part of being a ‘tourist’, i.e. of a person who leaves their familiar environment in order to encounter the world in a new and fresh manner, and that may, and usually will involve pleasant as well as unpleasant experiences. For example, my understanding of the institution of the American motel may be in need of an update such that continuous online availability becomes a required part of the guest role and physical presence is no longer a required part of the host role. But, overlaid on that local or regional tourist infrastructure now is a global infrastructure that increasingly organizes touristic travelling irrespective of local traditions and customs. And it does so in a manner which is typical of how platforms work and understand themselves, and therefore this experience may be milked for some insights into the nature of platforms.

A platform makes the services offered on it ‘swappable’. (For a long time, I had thought that I had been the originator of this clumsy term but, when writing this post, I discovered that it has already entered the English vocabulary, in the phrase ‘swappable connection’, which seems to express exactly what I mean to express.) I came to use this term as, over the years, I tried to explain the notion of ‘horizontal compatibility’ in my course on ‘Development of IT Standards’, offered every Winter term at RWTH Aachen University and attended mostly by computer science graduate students. Horizontal compatibility is the holy grail among standardization disciples. It means that instead of having just two components work with each other – normally called ‘interoperability’ or, among standardization nerds, ‘vertical compatibility’ – any component in one class of components can work with any component in another class of components. As a result, any component within one class can be substituted by any other component within that same class – swapped – and the replacement will still ‘interoperate’ with the component from another class which was previously ‘connected’ to the now replaced component. Think tires and cars, paper and copier, CD and CD player, etc. But recently a student in my class on standards challenged me when I said that it is possible to distinguish between ‘one-sided’ and ‘two-sided’ horizontal compatibility. The former means that components on both sides of the equation are swappable, the latter that swappability exists only on one side. His question was whether we should speak of a standard only in cases of two-sided horizontal compatibility or also when components are swappable only on one side. I replied that both would be instances of standards, thinking to myself that otherwise a lot of my pet examples for this course would no longer qualify as instances of standardization processes, such as Microsoft Windows. Actually, in the olden times, before Windows, there was DOS (for the younger of you dear readers: this meant ‘Disk Operating System’ since the OS was launched from a floppy disk, not to be confused with the modern meaning of this acronym in the phrase ‘DOS (Denial of Service) attack’) and it was swappable. There was MS-DOS, Microsoft’s version of it, but also alternatives such as DR-DOS (there is a good entry on the German Wikipedia page on this issue which, surprisingly, does not seem to have an English equivalent). However, as far as I know, Windows is no longer swappable (while, of course, it may be emulated, which is a way of making components swappable through adding a converter or adapter). Now, is Windows still a standard, or is it a platform? An intriguing thought which was actually suggested by the students who had challenged me on that issue.

The idea is that one-sided horizontal compatibility constitutes a platform and MS-Windows would be a perfect example. Microsoft can exploit positive network externalities – benefits that increase with the number of users – as application software becomes swappable. There is not just an immensely large inventory of applications for all kinds of purposes that run on Windows, but usually also many alternatives for each. But the other side of the equation, the operating system on which these applications run, is not. You have to have MS-Windows in order to tap into this rich ecosystem of applications. The term ‘platform’ may actually originate here, as the phrase of ‘running on’ invokes the metaphor of a solid and flat ground (‘platt’ in German means ‘flat’; ‘platform’ in German is spelled ‘Plattform’ accordingly).

To answer the question of my student, whether one-sided compatibility constitutes a platform, let’s first ask what the equivalent of Microsoft Windows would be in the hospitality industry. It is not Booking.com, but a hotel chain. Setting up an account with them will give you convenient access to a large number of hotels, which is probably the driving force behind the present race between Hilton and Marriott to become the largest hotel chain. If you choose to become a ‘member’ of that network, you become familiar with their way of operation. It is the same for any member of that hotel chain, similar to how Windows-compatible application software runs on every Windows computer. Your embodied knowledge, your practice, of how to interact with a member of that chain, possibly partly embedded in your computer settings and/or an app, is the one component and the way a hotel operates the other. Hotels are now swappable in that any hotel which belongs to the same chain can ‘inter-operate’ with you. But your practice is not swappable, it will not (smoothly) inter-operate with other hotels.

Now let’s factor Booking.com into this understanding. You might say that Booking.com is just the equivalent of a hotel chain writ large. But I would not agree. Any hotel and any hotel chain wanting to be listed on Booking.com has to comply with their process of how guests and hosts should interact. I actually recently discovered the problem that arises when a chain attempts to maintain its own process standards, when booking a holiday house in Denmark – for an excursion with students to study Denmark’s centralized medication record system – which belongs to a group (DanCenter). The process of booking and paying became incredibly messy for both sides. So, ignoring such cases, Booking.com makes both the practice of booking a hotel and the way a hotel operates swappable: the practice operates with any hotel listed on Booking.com and any listed hotel can inter-operate with any guest, whether or not that hotel belongs to a chain. Booking.com creates two-sided horizontal compatibility between hotel operations and booking practices.

So the answer to my student should be ‘no’: the difference between one-sided and two-sided horizontal compatibility does not account for the difference between a platform and a standard. A platform can also establish two-sided horizontal compatibility. Then what is the difference between a standard and a platform? My tentative answer is that a platform makes a business out of establishing one- or two-sided horizontal compatibility. Standards emerge through an evolutionary process, partly facilitated through collective bargaining, while platforms are the product of purposeful design aimed at maximizing exploitation of positive network externalities associated with horizontal compatibility. That may be a good thing when the evolutionary process fails to produce horizontal compatibility, e.g. because the institutional structure of an economy is such that it does not facilitate complementary collective bargaining processes. But the price paid is that most benefits, i.e. positive network externalities associated with horizontal compatibility, are privately appropriated. Hence, the alternative, to me, would not consist in fighting platforms, as the EU is currently doing, but in strengthening institutional structures that facilitate the evolutionary processes that produce horizontal compatibility. Sounds abstract and unrealistic? Yes, certainly. It is probably the modern incarnation of the Choice of Hercules. In our economic development, we have arrived at a crossroads regarding how we want to develop our infrastructures. Platforms offer the easy way.