(Published initially on August 26th, 2016)
“Technology” is a funny word, extremely malleable: we tend to recognize as technology only the most recent artifacts of human progress. In a more than purely metaphorical sense, technology is what appears in the world after we graduate. We call the likes of Google, Facebook and Amazon technology-based firms because the last great economic and social revolution hinged on the internet, but almost two centuries ago trains were all the novelty, and the technology firms of these age made huge fortunes covering the rolling plains of the Far West with a network of railroads among indian fights and bandit robberies, immortalized in John Wayne’s movies. Boilers over wheels replacing horses and carriages. Startups are technological because that allows to compete with the establishment.
What technology provides to a startup is, mainly, scalability. This is relevant not only since startups routinely lack the deep pockets of a fully grown business and must achieve a lot with very limited resources, but also because frequently a startup’s winning strategy relies on offering a superior product at an inferior price, yet marginally profitable, and quickly obtain the economies of scale that effectively create a de-facto monopoly. There are many ways to compete, but really few to win. The best way I ever heard to summarize it is in the candid words of Jack Welch, CEO of General Electric and professor at MIT: “do it first, do it cheaper, do it better or go home”.
We have a classic example in Google: Sergei Brin and Larry Page created the company around PageRank, their thesis’ algorithm to automatically classify a website’s relevance and thus facilitate internet search. Infinitely more scalable that the technology used by the incumbents of the time: rooms full of clerks constantly browsing and manually classifying the information.
Interestingly in 1998 Google offered to sell their technology for 1 million dollars because the founders preferred to continue studies at Stanford, but both AltaVista and Yahoo declined the outrageous offer. Yahoo had yet another opportunity to acquire Google in 2002, but the agreement fell apart after weeks of hard negotiations because Terry Semel, then Yahoo’s CEO, faltered at the clearly exaggerated 5 billion dollar price tag.
As I write these lines at the beginning of September, 2016, Google and its parent organization Alphabet have a market capitalization in excess of 500 billion dollars. This is a humongous level of value creation. Meanwhile, in July this same year it was announced that Yahoo reached an agreement to sell its internet properties to Verizon for 4.83 billion dollars. It is worth remembering that in 2008 Microsoft bid 44.6 billion dollars for all of Yahoo’s businesses. That, dear readers, is value destruction.
An algorithm defeated an entire industry, and this fact has been often repeated since distributed computation is at the spearhead of technology. AirBnb, Uber y Taskrabbit are today’s reference companies worldwide, followed by a myriad of startups pursuing more specific market opportunities in sharing economy, crowdsourcing and crowdfunding. Through digital marketplaces that automatize the exchange of services among individuals, again technology allows reaching that scalability that makes possible the disruption of an established industry.
Then, what does future hold for us? We already know that predictions are only reliable when performed backwards, but it is likely that the next great technological revolution will come from the world of artificial intelligence (AI), the set of methodologies that strive to replicate human cognitive processes in machines.
Here we have an interesting blind spot: as we mentioned above technology is all that’s new in the world, but AI is veiled with the magical halo with which we tend to dress everything that surrounds matters of the human mind. Rodney Brooks, professor of AI at MIT, refers to it as the AI effect: “When we finally manage to understand a piece we say ‘Oh! That is just computation’ and we leave as AI only the unsolved problems”. We should be aware that elements of AI have been for years “infiltrating” our daily experience, often in rather subtle ways: the simple ability to automatically tag friends in pictures, the kind interactivity of a virtual assistant, the cunning behavior of an enemy inside of one of our favorite computer games.
Now both technical advances in fields like deep learning and the availability of massive amounts of data with varying levels of structure (what is usually known as big data) will allow us to perform automated analysis in areas of human activity that just a few years earlier required a high level of domain-specific knowledge and were thus confined to the hardly scalable niches of the expert and the consulting boutique.
An example that is close home: smart algorithms endow us at Ágora EAFI to automate the tedious task of analyzing and classifying fundamental company information not about a single firm, but thousands of them, unearthing relevant patterns without the need for hiring rooms full of technical analysts, the associated costs and headaches. Another example that I find really interesting is Descartes Labs, a startup based in USA that employs machine learning algorithms applied over pictures taken via low orbit satellites in order to more precisely predict crop yields. While the method employed by the department of agriculture (USDA), consisting of asking the farmers around the country, produces one estimate each month, Descartes Labs updates its own each two days. Imagine using that privileged information to speculate in the market with futures. Big deal.
And beyond? If the advances in biotechnology are up to our expectations, then maybe the algorithm will become flesh. But please stop me here before my philosophical knack runs wild and I bore you even more. Cheers and talk again soon!