Big Data is still only Small Data

Big data: "Big Brother" resonates here, which in view of the NSA revelations is perfectly suited to stir up surveillance fears among the public. The term, which a few years ago was only familiar in specialist circles, is now increasingly ubiquitous as a buzzword and thus defies a precise definition. Basically, it is about deriving benefits from the rapidly increasing amount of data that is being collected everywhere. For the individual, for business, for society, for the planet. It is clear to everyone that there is more and more data. But it is not the flood of information on the web and in social media that is primarily responsible for this. It is mainly machine-generated data that is growing like an avalanche. In telecommunications connections, web access, goods registration in logistics, in the financial industry. Hardly any sector of the economy is exempt from this. And this development is only just beginning: with the Internet of Things, billions and billions of sensors will ultimately be constantly supplying data. Seen in this light, what is referred to today as big data is only a small precursor of what is yet to come. In his book "The Zero Marginal Cost Society", social philosopher and economics professor Jeremy Rifkin diagnoses the emergence of a new economic system (the collaborative commons). Because marginal costs are tending more and more towards zero due to digitalization and networking, capitalism will no longer be worthwhile in the future.

Old hands know: we've seen it all before. Ten or fifteen years ago, it was called CRM (customer relationship management) or came in the form of business intelligence (BI). Back then, too, mountains of customer data were compiled and analyzed. It was expensive and complicated and did not live up to expectations. This observation is sometimes correct, but it is wrong to assume that big data is also a hype that is top today and flop tomorrow (and can be safely ignored and sat out). Just as videotex was a precursor to the Internet and paved the way for awareness of corresponding processes, big data is the next wave after CRM technology. Back then, data was stored in a so-called data warehouse; today, the term "data lake" is used. Whereas back then data was analyzed using data mining, these processes are now known as big analytics (and are often included in the term big data). The promises (competitive advantages) were the same and the price for the necessary systems (high) too. Then as now, there was a lack of specialists who could handle the new technology skillfully. Added to this (at least in Europe) are fears about data protection and, as a result, (anticipated or real) restrictions that can hinder the benefits of big data.

Introduce company-wide

The comparison with CRM systems is useful for learning from the mistakes of the past. Because CRM (and big data) required new technical infrastructures, such projects were often managed by IT. Sooner or later, the corresponding systems were reported to be working and the investments were justified, even though the operational business units did not see any benefits. If companies want to launch a big data initiative today, all areas must be involved from the outset. Delivery drivers, for example, need to understand why their journeys are being recorded and evaluated in real time (e.g. to save time and money by choosing a better route), and they will learn to explain certain incidents in such a way that this data can be better evaluated in the future. In marketing and sales, it is important to understand how analysis results are obtained and to what extent they are useful for business decisions or merely provide vague indications. In production, the use of sensors is constantly being improved and in order to make processes more efficient, production specialists and data analysts must be able to discuss their evaluation on an equal footing. Science fiction illusions must be put aside: No matter how many computers are used for Big Data, they still do not know anything (in the human sense). There is therefore a great danger of assigning the wrong meaning to correlations that can be found in the data volumes (especially if the results are presented as attractive graphs) and moving in the wrong direction (because everything was so expensive, it must be right ...). It is therefore not only important to be able to store and manage the data, but also to draw the right conclusions from it. Last year, management consultants Bain & Company analyzed the experiences of more than 400 companies that got off to an early start. The frontrunners are characterized by these advantages:

- they are twice as likely to rank in the top quartile for their financial performance within their sector
- it is five times more likely that they can make decisions much faster than competitors
- it is three times more likely that they will be able to execute decisions as intended
- it is twice as common for them to use data very frequently in their decisions.

The management consultants therefore conclude: "Many of these leading companies are currently pulling ahead of their competitors, leaving them lagging behind and struggling to catch up."

America ahead, the EU behind - and Switzerland?

But it's not just companies that have to fight to keep up with big data, countries do too. And it looks like Switzerland is not among the leaders here - although the conditions would be good. The USA, where the Obama government provided a budget of 200 million dollars for the Big Data Research Development Initiative in spring 2012, which will benefit various American authorities, is ahead of the game. This is entirely consistent: After all, Obama's last election campaign was the first big data campaign in history. The Democratic Party's election strategists are using big data analytics and behavioral research to target voters individually. In addition, there are numerous other extensive research programs and applications in companies in the land of opportunity - not least in the area of online advertising, where real-time bidding (RTB) is used to determine placements and their prices based on large volumes of data on surfing and consumer behaviour.

The EU has also woken up to big data and at the beginning of July, the EU Commission's Directorate-General for Communications Networks, Content & Technology (CONNECT) presented a communication (well worth reading) for a "thriving data-driven economy". This is an action plan for the EU to keep pace with the global data economy, in which the EU's research and development funding program, Horizon 2020, plays a significant role. It is worth noting this section: "This global trend holds enormous potential for various areas, from health and food security to climate policy, resource efficiency, energy, smart transportation systems and smart cities. Europe cannot afford to miss this opportunity. But compared to the US, the digital economy in Europe has been slow to embrace the data revolution. There is also a lack of comparable industrial capacities in this country. Research and innovation funding in the field of data is insufficient in the EU and the relevant activities are also largely uncoordinated. There is a lack of data experts capable of translating technical advances into concrete business opportunities. The complexity of the current legal framework, combined with the insufficient accessibility of large data sets and corresponding basic infrastructures, creates barriers to entry for SMEs and hinders innovation." The EU Commission summarizes its demands in these points:

- Support "lighthouse" data initiatives that can help increase competitiveness, improve the quality of public services and improve the lives of citizens. Possible areas include healthcare (personalized medicines), integrated transport and logistics management for entire regions, food supply chain management by tracking food from farm to fork

- develop the basic technologies, infrastructures and skills, especially for the benefit of SMEs

- make their public data and research data infrastructures available for shared use on a large scale and also use and further develop them

- focus public research and innovation on existing technological, legal and other bottlenecks

- ensure that the legal framework and policy requirements, e.g. for interoperability, data protection, security and intellectual property, are data-friendly in order to increase legal certainty for companies and strengthen consumer confidence in data technology

- swiftly complete the legislative process to reform the EU data protection framework and network and information security and support the exchange and cooperation between the relevant law enforcement authorities (e.g. in relation to data protection, consumer protection and network security)

- accelerate the digitization of public administrations and public services to increase their efficiency, and

- help to ensure that the results of new data technology reach the market by awarding public contracts.

Such a strategically oriented big data initiative is lacking in Switzerland. The State Secretariat for Economic Affairs (SECO), which is (co-)responsible for this, at least comes to the conclusion in the 2014 E-Economy Report: "The phenomenon of big data should be analyzed in detail."

Christoph J. Walther

Tyler Vigen is a law student at Harvard Law School and describes himself as a statistical provocateur. On his website (tylervigen.com), he uses numerous examples to show that a correlation between two variables does not necessarily imply causality - although the temptation to do so is often great, especially when presented in a visually attractive way. In the search for optimization potential, many of the current big data algorithms come across such apparent correlations that are not based on causal relationships. This is one of the great dangers of big data: anyone who relies on algorithms in good faith can easily come to the wrong conclusions. This can not only lead to costly wrong decisions, but also to discrimination and other disadvantages.