In the series “algorithm regulation”, we discuss the implications of the growing importance of technological algorithms as a means of regulation in the digital realm.
Today I stumbled via twitter upon the website “Google Algorithm Change History” that chronologically documents all changes of the core search algorithm publicly announced by Google. The most striking feature of the site is the sheer number of changes:
Each year, Google changes its search algorithm up to 500 – 600 times. While most of these changes are minor, every few months Google rolls out a “major” algorithmic update that affect search results in significant ways.
In other words, it does not make sense any more to speak of “the Google algorithm” because there is not an algorithm but there are algorithm-related practices. In line with the practice turn in contemporary social theory (see Schatzki et al. 2001) and similar to perspectives such as strategy-as-a-practice, we might require a practice perspective on algorithms to better understand how algorithm regulation works.
When looking at the frequent – not to say constant – changes in Google’s search algorithm, it also becomes obvious how misleading regular comparisons with the Coca-Cola formula such as the following in a Wall Street Journal blog are:
Google is very cagey about its search algorithm, which is as key to its success as Coke’s formula is to Coca-Cola.
The algorithm of Google search is not like a static formula and therefore it should not be treated as a trade-secret either. Actually, if the search algorithm where a mere formula, we would see much more competition in search. Google is practicing algorithmic search and it is these continuous changes, which mostly rest on access to unimaginably big data sets of search and usage practices, that are difficult to imitate for competitors.
With regard to the issue of algorithm regulation, a practice perspective sensitizes for phenomena such as regulatory drift. In a paper on transnational copyright regulation, Sigrid Quack and myself describe regulatory drift as “changes in meaning and interpretation, which result from continuous (re-)application of certain legal rules” (see also Ortmann 2010). In the context of algorithms, the term might refer to the sum of continuous revision and (seemingly) minor adaptation practices, which in the end lead to substantial and partly unintended changes in regulatory outcomes.
(leonhard)
4 comments
Comments feed for this article
April 13, 2013 at 15:33
Algorithm Regulation #5: Regulating Algorithms |
[…] results is difficult. Google is not only constantly changing its search algorithm (see “Algorithm Regulation #4: Algorithm as a Practice“) but also increasingly personalizing search results; both these characteristics of […]
June 29, 2014 at 14:18
Algorithm Regulation #7: Experimenting with Algorithms for Business and Research |
[…] tend to agree. Algorithms are neither neutral nor static; rather, we can even speak of algorithms as a practice, as is evidenced by Google changing its search algorithm 500 to 600 times a year. And at the heart […]
May 8, 2016 at 22:45
Rezension: „Kultur der Digitalität“ von Felix Stalder | netzpolitik.org
[…] (S. 185) erlaubt. Algorithmen sind allerdings keineswegs statische Rechenfolgen, sondern kontinuierlich adaptierte algorithmische Praktiken, die individuelle Nutzungshistorien mitberücksichtigen und so die Welt „für jeden User […]
April 18, 2019 at 16:50
Neues aus dem Fernsehrat (37): Wir müssen endlich über Archive reden – netzpolitik.org
[…] Metadaten oder die Kategorisierung und Beschreibung von Inhalten sind demnach zunehmend Teil von algorithmischen Praktiken. Gerade für öffentlich-rechtliche Empfehlungsalgorithmen, die sich eben nicht nur an Klickzahlen, […]