Summer is finally here and the nights are long, which gives us plenty of time to think about the fundamental questions of life. That’s why the July issue of SEO News examines not just the forthcoming Google updates, but also the cognition game show and the future pecking order on our planet.
1) Achieve good rankings quickly and securely
Once again, Google is focusing on the convenience and security of Internet users. The company (which in its own words aims to do no evil) is launching not one, but two updates in July – whose effects will be of equal benefit to Internet users and website operators alike. Both of these changes were announced a long time ago and have already been partially implemented. The first change will see the loading speed of mobile websites become an official ranking factor. Loading speed is already listed as a quality criterion in Google’s top 10 basic rules for website quality; however, it has taken a very long time for it to become a genuine ranking factor. The change was originally introduced based on studies showing that slow-loading websites experienced direct impacts on their clickthrough and conversion rates, and the speed argument was also repeated like a mantra by Google representatives at various search conferences during the 2018 season. The subsequent introduction of the Mobile First Index (see our report here) means that the rule has now been made official for mobile sites too. Google recommends that website operators analyse their domains using Google’s own “Page Speed Report” and “Lighthouse” tools and make the necessary changes for mobile websites. Alongside its speed update, Google is also getting serious in July with its announcement that websites which are not converted to the encrypted HTTPS protocol before the deadline will be marked as “not secure” on Chrome. This change also marks the end point of a campaign that was launched over two years ago in 2016, when Google began its awareness-raising work with a small ranking boost for secure websites. Google has described that measure as a success, with the company stating that around 68 per cent of all Chrome traffic on Android and Windows now occurs over HTTPS – and there is plenty of scope for that percentage to grow. The fact that Google is leveraging its market power to implement technical standards with the aim of improving the user experience is a step in the right direction. Many companies were only prepared to invest in faster technology or secure licences when threatened with reductions in traffic or sales. In order to prepare for future developments, it is advisable to keep an eye on new technologies such as AMP (Accelerated Mobile Pages), mobile checkout processes, and pre-rendering frameworks that allow content to be pre-loaded. These innovations can help you keep pace, especially when it comes to improving user perceptions of loading rates on all platforms.
2) Life is one big game show
This bit will be tricky for those of you who didn’t pay attention in maths. Remember that moment back at school, somewhere between integral calculus and stochastic processes, when you belatedly realised that you’d completely lost the plot? Well, in the age of algorithms that will come back to haunt you – especially if you work in online marketing. In everyday terms, an algorithm is nothing more than a carefully ordered chain of decisions designed to solve a problem in a structured way. The crucial innovation in recent years is the advent of artificial intelligence and machine learning. Nowadays, the individual links in the algorithmic chain are no longer assembled by people, but by programs. When you ask a search engine a question, the information is taken in, its core information object (the entity) and intention are identified by means of semantic analysis, and the most empirically appropriate result (the ranking) is returned in the correct context (e.g. local and mobile). However, a group of seven Google engineers presented a research project at the ICLR Conference in Vancouver that turns the question/answer principle on its head. For their project, the researchers used tasks taken from the popular US game show “Jeopardy”. On this show (first aired in 1964), contestants are required to provide the right questions in response to complex answers. In their study, the Google engineers exploited the fact that Jeopardy tasks involve information deficits and uncertainties that can only be resolved by formulating the right question. In other words, the question needs to be adapted until the information provided in the answer makes sense in its specific combination and context. The human brain performs this task in a matter of seconds, and is able to draw upon a comprehensive range of intellectual and social resources as it does so. However, if you ask a Jeopardy question (such as “Like the Bible, this Islamic scripture was banned in the Soviet Union between 1926 and 1956”) to a search engine, you will not receive an appropriate answer. Google returns a Wikipedia article about the Soviet Union, meaning that it interprets this search term as an entity or a core information object, and thus falls short. Microsoft’s search engine Bing comes a little closer to the obvious answer from a human perspective (“What is the Koran?”), but is likewise unable to deliver a satisfactory result. This little trick involving Jeopardy questions makes clear what the biggest problem is for search engines (even though it is marketed as one of the main markers of quality for modern search systems): how to accurately recognise the intention behind each search query. The idea is that what SEO professionals in companies and agencies currently work hard to develop should be reliably automated by the search engines themselves. In order to achieve this, the Google researchers developed a machine-learning system that reformulates possible answers to the Jeopardy question into many different versions before passing these on to the core algorithm itself. In step two, the answers obtained are then aggregated and reconciled with the initial questions. The results are only presented to the user once these two intermediate steps are complete. The self-learning algorithm then receives feedback on whether its answer was right or wrong. The AI system was subsequently trained using this method and with the help of a large data set. As a result of this training, the system learned how to independently GENERATE complex questions in response to familiar answers. This milestone goes far beyond simply UNDERSTANDING search queries, which are growing increasingly complex under the influence of voice and visual search. Although this study was carried out by Google, we can assume that Microsoft, Yandex and Baidu are also working on equivalent technologies designed to further automate the recognition of search terms and to automatically generate complex, personalised content in the not-too-distant future. At present, however, it is impossible to gauge what effects this might have on the diversity and transparency of the Internet.
3) Google Assistant sets the tone
While we’re on the subject of automatic content generation, we also have an update on Google’s uncanny presentation of two phone calls between the Google Assistant and the working population. Back in May, the search engine giant from Mountain View presented a video at its “IO” developer conference in which an AI extension to the Google Assistant named “Duplex” booked an appointment at a hairdresser’s and a table in a restaurant entirely on its own, all while perfectly imitating human speech. The human participants in those conversations were apparently unable to recognise that they were interacting with a machine while they went about their work. Close collaboration with robots and AI systems has long been familiar to industrial workers in the Western world, but now this development is also moving into the service economy, and therefore into our day-to-day lives. At first glance, the Google scenario was astonishing and convincing; however, the unnerving initial impression was swiftly followed by a number of pressing questions. In particular, the fact that Duplex failed to identify itself as a machine to its human conversation partners was the subject of considerable debate. Google has since responded and published a new video in which the Google Assistant identifies itself at the start of the conversation and states that the call will be recorded for quality control purposes – similar to a recorded message in a call centre. Taking a more detached view, however, one wonders whether this responsiveness on the part of the artificial intelligence is actually completely superfluous. The restaurant employee in the video follows the Google Assistant’s instructions obediently, as if he is talking to a human being – there is no difference whatsoever. In search marketing, we attempt to further our own interests by reflecting the intentions of target groups and consumers in the content produced by search engines (the results pages). In voice search, we issue commands to a machine – and a number of years will pass before we learn how that will change us. And in Google’s future scenario of an invisible, omnipresent and convenient system that allows users to organise themselves and solve problems, the human simultaneously becomes both the subject and the object of the technology. Our data was used to create, feed and train the system, and so we may briefly feel ourselves to be its masters; however, given the current state of affairs, we can and should seriously question whether we will recognise the point of no return once the balance finally tips.