Epistemic Injustice in Technical Interviews

What is epistemic injustice and how does it manifest in technical interviews?

Epistemic injustice occurs when someone is wronged in their capacity as a knower. In technical interviews, this happens when a candidate’s knowledge is discounted because of who they are (testimonial injustice) or when the interview format cannot recognize forms of knowledge the candidate possesses (hermeneutical injustice).

I interviewed a candidate who had built and maintained the data infrastructure for a nonprofit serving 400,000 beneficiaries. They managed a 3-person team, handled compliance with 5 regulatory frameworks, and processed 2 million records monthly on a $3,000/month cloud budget. They could not solve the whiteboard algorithm problem. The algorithm problem tested abstract problem-solving divorced from context. The candidate’s actual work demonstrated applied problem-solving under severe constraints. The interview valued the former and discounted the latter.

This is hermeneutical injustice. The interview framework had no way to recognize the knowledge the candidate possessed. Constraint-based engineering, stakeholder management, regulatory navigation, these are real forms of technical knowledge. The whiteboard has no category for them. The candidate was not lacking knowledge. The interview was lacking the concepts to recognize the knowledge they had.

How does testimonial injustice operate in technical interviews?

Through credibility deficits: the unconscious tendency to give less weight to the testimony of people from groups perceived as less technically capable. This operates even when interviewers believe themselves to be objective.

I conducted an experiment within my own hiring process. I anonymized resume reviews for 6 months, removing names, universities, and company names. The callback rate for candidates from non-traditional backgrounds (bootcamps, self-taught, non-CS degrees) increased by 34%. The resumes had not changed. The identifiers had. The previous callback rates reflected not the quality of the candidates but the credibility assigned to their backgrounds.

Fricker argued that testimonial injustice is structural, not individual. Individual interviewers may be perfectly well-intentioned. But the structures they operate within (resume screening, whiteboard coding, culture-fit evaluations) embed credibility deficits that no individual intention can overcome. As I explored in imposter syndrome as Socratic wisdom, the people who doubt their belonging often have the clearest perception of the structures that question it.

How can interview processes be redesigned to reduce epistemic injustice?

By expanding the interview’s epistemic framework: recognizing more forms of knowledge, evaluating candidates on tasks that resemble actual work, and structuring evaluation to resist credibility biases.

• Work sample tests over algorithm puzzles: Ask candidates to solve problems that resemble the actual work. Review real code. Debug a real issue. Design a real system. These tasks evaluate applied knowledge, which is what the job requires.
• Structured interviews with rubrics: Unstructured interviews maximize the influence of bias. Structured interviews with explicit evaluation criteria reduce it. Every question should map to a specific competency, and every competency should be evaluated against a defined scale.
• Anonymized first rounds: Remove identity signals from early screening. Evaluate work before evaluating background.
• Expand the knowledge taxonomy: Recognize constraint-based engineering, stakeholder communication, system maintenance, and self-directed learning as technical competencies. The whiteboard tests one form of knowledge. The job requires many forms.

Why is this a philosophical problem, not just a diversity problem?

Because epistemic injustice is about what counts as knowledge and who counts as a knower. These are epistemological questions, and answering them requires philosophical work, not just procedural changes.

The diversity framing asks: “How do we include more people?” The epistemic justice framing asks: “What forms of knowledge are we failing to recognize, and whose knowledge are we systematically discounting?” The second question is deeper because it challenges the interview’s epistemology, not just its demographics. A diverse candidate pool evaluated through an epistemically narrow interview produces diversity without inclusion.

According to Fricker’s analysis, epistemic injustice is both a moral wrong (the person is harmed) and an epistemic loss (the organization loses access to knowledge it needs). Technical interviews that systematically undervalue certain forms of knowledge do not just harm the excluded candidates. They harm the organizations that fail to recognize what those candidates know. The connection to epistemic humility is direct: if you are not humble about what forms of knowledge matter, you will not recognize the knowledge that walks through your door.

“The interview does not just test knowledge. It defines what counts as knowledge. That definition is a site of power, and it can be a site of injustice.”

Technical interviews gate access to careers worth $120,000 per year. They operate as epistemic gatekeepers, determining not just who gets hired but what counts as competence. When those gates are structured around a narrow definition of technical knowledge, they produce a workforce that reflects the definition, not the actual range of knowledge the work requires. Fricker’s framework reveals that the problem is not just who we interview. It is what we are prepared to recognize as knowledge. Expanding that recognition is philosophical work. It is also the most practical thing an engineering organization can do to improve the quality of its talent pipeline.