Mar 05, 2021
6 mins read
Before we embark on this small journey, I want to express my gratitude to the original researchers of this paper for paving the way for ethical product development. Though I feel 'weird' writing reviews on other peoples work as at the moment there is definitely more for me to learn on the topics the paper discussed, nevertheless I hope this review will spark your interest to read the original paper and take what is written in there into consideration for your own work.
Read the original research paper here by Humphrey O. Obie, Waqar Hussain, Xin Xia, John Grundy, Li Li, Burak Turhan, Jon Whittle, Mojtaba Shahin
Presented study explores violations of human values in apps through analyzing their reviews and then categorizing findings accordingly to human value theory model. The authors underpin differences between micro- and macro-values of an individual and through their study aim to coin in the importance of taking in consideration alignment of all stakeholders (brand, engineers, developers, users) values in the product creation. They argue that understanding the interplay between human values and mobile technologies is crucial as it directly affects the fundamentals of the society. To support this argument, the paper presents real-world case study when an error in an app caused loss of lives in Australia, which alone seemed as a very strong point. Seeing how technology is becoming more present in the daily lives of our society, and everyone individually feeling the effects online world has IRL to one extent or another, it is then valid to support the need for more exploration into ethical software development. It is equally important to involve the values of the end users in the developing process, too.
The authors choose an efficient approach of analyzing large volume of mined text through natural language processing (NLP) technique combining sentiment analysis, app feature extraction, and a human values dictionary of keywords combined with Schwartz model. However, as researchers have mentioned themselves, this technique had its limitation as it could only involve reviews on English language, and I would argue this to be a weak point of the paper. If we take in consideration that most apps used in the study are internationally recognized, we must then understand that through exclusion of participants with other language knowledge this study is inconclusive. Foremost, there is a high probability inclusion of other languages will increase the end results of 26.5% [of app reviews showing violation of values].
Secondly, with respect to the fact that values are variable in nature, it is also important to take in consideration their inevitable contradiction based on culture, geographical location, and historical context. If we refer to Schwartz model, categorized values are placed into individual sections, and some of these sections will either be complementary or contradictory to each other. For example, opposite sections that will be polar are conformity versus open-mindness, or individualism versus collectivism, but at the same time open-mindness and individualism will be complimentary, even though they are placed in different sections. Here must happen exploration of the question whether there must be a standard framework created for developing ethical software. Researchers must understand that that would imply fundamental global societal shifts in values towards those that would be deemed more widely acceptable. History witnessed similar events from both good (eg. 70s human rights movement) and bad (eg. Third Reich regime) perspectives and numerous scholars have explored the psycho-social l impact global events had on then present and future generations. To explain further, if in the past, considering the war times, collectivism had been a logical value in place as a need for survival, the generation that came after, during peace times, had more opportunities for self-exploration. As such, more space opened for individualism paired with the rise of globalization. At the same time technological progress created new environment for these generations to live in. Through rapid digitization we have witnessed people of different backgrounds and, therefore, values come together in an unexplored environment. As the researchers highlighted, it is indeed important to align values of all stakeholders, including end users, in the creation of the product for the benefit, but since developers are now working in this new environment, who and how should control values within in?
Reasonably, we must then look at values more as guiding principles of human behavior. The theory proposed by Gouveia et al. “Functional theory of human values: Testing its content and structure hypothesis” creates a new dimension to look at Schwartz model and ways to apply it in research. Subdividing values to two groups as guides of actions and expression of needs, Gouveia et al. have pinpointed a scale of goals ranging between personal (the individual by itself) to social (the individual in the community) with central (the general purpose of life) being the point of reference. The researchers argue that individual values are determined by the need for survival (life as source of threat) and thriving (life as source of opportunities) and therefore when all basic human needs are met and there is no danger around, the individual will shift values towards understanding general purpose of life. Firstly, this paper provides a new perspective at how to align stakeholders’ values more efficiently, and secondly, it may be offering an ethical compass for software development. Even though the work of Obie et al. fails to paint a bigger picture of the violations based of each participant personal needs and goals, it still provides interesting evidence of what types of values are being violated. The study had pinpointed most violated values to be benevolence and self-direction, and conformity and tradition as least and this finding may cause speculation whether there is a bias present in the digital world as of now. If we compare these findings to the framework of Goudvie et al., we can see how values that align with social goals (tradition and conformity) are overpowering central (benevolence) and personal (self-direction) goals.
On the other hand, since the presented study was conducted on a large group of people from various backgrounds, it fails to paint a more precise picture of the end user of a particular platform and which values they hold. For example, the study looks at such apps as TikTok and Minecraft that are heavily populated by Gen Z, but also at Tripadvisor and Telegram which are predominantly used by Gen Y and Gen X. Each generation, as was mentioned above, is led by its own set of goals and values based on their upbringing environments. This means that, for example, those from Gen X are more likely to accept importance of tradition and conformity in an app, and self-direction to be less important. This will be less likely for Gen Y and Z. Therefore, though a generalized study provides insights into violations of values in apps, such method of product evaluation cannot be conclusive without taking in consideration which values are dominant for the end user.