Do Startup Employees Earn More in the Long Run? – I would have thought that startup employees would earn more, but this study in the Danish context found that startup employees ended up with less 17% less salary over the next 10 years. They attribute these to the following factors: selection effects of people who go to startups, the high failure rates causing unemployment periods and inability to return to larger employers.

Ten computer codes that transformed science – fascinating article on how computing has transformed the scientific process.

The Sequence Effect in Panel Decisions: Evidence from the Evaluation of Research and Development Projects – the book “When” summarizes how timing affects how people evaluate things like who would get a scholarship or who will win a competition. If you are evaluating 10 things and you already said yes three times in a row, you would probably tend to say no to the next one just to “balance” things out. Even experts are not immune to the sequence effect as shown by this study.

Artificial intelligence and management: The automation–augmentation paradox – When technologists discuss AI, they talk about how many jobs would be replaced due to automation. When economists discuss AI, they typically focus on how such transformational technologies are not new and that as history suggests, it would probably just augment the performance of workers. This article talks about how these distinction between automation and augmentation is false as these two are interrelated.

An empirical meta-analysis of the life sciences linked open data on the web – talks about the Life Sciences Linked Open Data cloud which enables researchers to seamlessly discover and integrate data from multiple data sources that had been openly available online but had previously been disconnected from one another.

Evaluating impact from research: A methodological framework – proposes a typology of research impact evaluation methods. They identify the following designs: experimental and statistical methods, systems analysis, textual, oral and arts-based methods, indicator-based approaches, evidence synthesis approaches.

Appraising research policy instrument mixes: a multicriteria mapping study in six European countries of diagnostic innovation to manage
antimicrobial resistance – before I started my postdoc, I wrote a grant for a pharmaceutical research consortium to address AMR. This study is interesting just to check out what their method called multicriteria mapping is. As for the content, it is interesting as it explores policymakers’ preferences and uncertainties with various policy interventions against AMR. The policy options they explore were categorized in the following:

• incentivizing diagnostic firms with financial rewards
• funding R&D
• coordinating stakeholders to make it easier to bring new tests to market
• the government provisioning resources to lead R&D and testing themselves
• incentivizing healthcare providers to use tests more appropriately
• establishing IP regimes to support the development of tests based on demand

Deep Learning applications for COVID-19 – explores how deep learning is used in COVID. They categorize the application to four main areas: computer vision (ie. to analyze medical images and robotics), life sciences (ie. for drug repurposing and protein structure prediction), epidemiology (ie. for forecasting and contact tracing) and natural language processing (ie. literature mining).

Psychological factors influencing technology adoption: A case study from the oil and gas industry – introduces the Psychological Technology Adoption Framework (P-TAF) which categorizes various factors that facilitate or hinder adoption: personality (innovativeness, risk aversion), motivation (personal incentives, fear of technology failure), attitude (technology attitudes, trust), cognitive (risk perception, technical knowledge, certainty perception, previous experiences), social (social influence, subjective norms) and organizational (leadership, collaboration culture, technology adoption culture).

Legitimation of a heterogeneous market category through covert prototype differentiation – introduces the idea of covert prototype differentiation where entrepreneurs communicate how the category that they are in is united while at the same time creating prototype variants that attract entrants to their own camp.

Happy new year everyone! Here are my interesting reads from the holidays up to today.

Developing a unified definition of digital transformation – studies that aim to define a certain concept are always fascinating. In this study, they review the digital transformation literature and complement it with a survey of practitioners and academics to get a unified definition of DT:

A fundamental change process, enabled by the innovative use of digital technologies accompanied by the strategic leverage of key resources and capabilities, aiming to radically improve an entity* and redefine its value proposition for its stakeholders.

Organizing for innovation: a contingency view on innovative team configuration – study on how a team’s expertise should be tailored according to the domain they are trying to innovate in. The researchers explore 4 domains varying by modularity and breadth of application – MRI, RFID, stem cells and nanotubes. Modular domains like MRI and RFID do not need a lot of overlap in knowledge across their inventors. Domains with broad applications benefit from teams with wide knowledge breadth as well.

The front end in radical process innovation projects: Sources of knowledge problems and coping mechanisms – It’s worth the read even just for the review on the difference between complexity, uncertainty, equivocality and ambiguity. These are terms that are thrown around interchangeably and this article builds on previous works to explain them more clearly:

• Uncertainty – refers to the lack of sufficient information needed to get to the desired outcome. Put simply, not knowing the answers.
• Complexity – related to the number of parts in a system and the difficulties in predicting the interactions among them. In other words, not knowing how to find the answers to the question.
• Ambiguity – the inability to interpret. In other words, not knowing what questions to ask in the first place.
• Equivocality – potential for multiple interpretations, even when information is available. Put simply, having multiple answers.

Artificial intelligence in drug discovery: what is realistic, what are illusions? Part 1: Ways to make an impact, and why we are not there yet – despite moving away from pharma research, I still enjoy reading news about the industry. The summary of this article is the classic line garbage in, garbage out. The conclusion perfectly illustrates this: “with our current ways of generating and utilizing data, we are unlikely to achieve the significantly better decisions that are required to make drug discovery more successful… we need to understand what to measure, and how to measure it… Only once these data are available for AI approaches can the field be expected to make real progress.”

Impact for whom? Mapping the users of public research with lexicon‑based text mining – When I moved to the social sciences, the impact of my research became much more difficult to define and perceive. This study maps the various communities that such research can have impact to. Found the visualizations fascinating.

Who Contributes Knowledge? Core-Periphery Tension in Online
Innovation Communities – explores what makes a contribution highly valued in the forum Stack Exchange. They talk about social embeddedness which refers to how connected socially an individual is to their community and epistemic marginality which refers to how different an individual’s knowledge base is compared to their community. Those who are both socially embedded and epistemically marginal are able to contribute highly valued insights as these insights are generally novel compared to the community and at the same time, they are able to communicate well their ideas to their community.

Recommender systems in the healthcare domain: state-of-the-art and research issues – always interesting to see the overlap across healthcare and AI. This review looks at the application of recommender systems in food, drug and exercise recommendation, health status prediction and health care professional search.

It did not work? Unlearn and try again – Unlearning success and failure beliefs in changing environments – it’s important to learn from your mistakes, but it’s also important to unlearn the false beliefs you have formed from previous mistakes. Cool insight considering that people generally focus on learning and not on unlearning.

Experience base, strategy-by-doing and new product performance – When we think of strategy, we usually think of getting a whiteboard and brainstorming on the best plan forward. In contrary, there is strategy-by-doing which is defined as “actions of iterative search for what works in the market.”