Term 7 · Module 5 of 8

Expertise, Authority, and Value

Critical Thinking

Practise this module in Claude, ChatGPT or Gemini

Set it up free
Mind map

Expertise, Trust, and Credibility

Who do we believe, and why?

Two people both distrust the mainstream media, but for very different reasons.

Maya Manan
History Has barely encountered news; her feeds, friends and habits never surfaced it Has seen plenty of news, but his close-knit group taught him long ago that outsiders cannot be trusted
Nobody told her / him Nobody told her to mistrust outsiders Outsiders are discredited in advance
Shown the missing evidence May change her mind Digs in deeper
Structure Epistemic bubble Echo chamber
  • An epistemic bubble: relevant voices are left out, by accident, algorithm, friend group or habit. It is fragile; add the missing voices or change the algorithm and it pops.
  • An echo chamber: outside voices are present but discredited in advance because only the in-group is accepted as telling the truth. It is a constructed structure that resists change.

Main claim: you cannot understand expertise by looking at experts alone. You must look at the networks of trust around us, because they decide whose expertise we even want to hear. (Concepts drawn from the pre-read by Nguyen.)

Four questions: what expertise is, its kinds, how trust and credibility work, and how to question experts without ending up like Manan.

What makes someone an expert? Four signals

Common answers: knowledge, track record, credentials, peer recognition. Each is a signal, and each can mislead.

Signal Why it helps Why it can mislead
Knowledge More knowledge on a topic suggests expertise Knowing the rules of chess does not beat a 10-year-old who plays constantly. Studies of chess masters found experts do not consider more options than beginners; they see options differently, recognising patterns others cannot perceive. Expertise is knowing which facts matter, not just having facts
Track record The strongest evidence: good results that are hard to fake Works only when reality gives clear, quick feedback. A chess player knows within moves that they blundered; a political pundit may wait months to learn whether an election prediction was right, by which time people forget. Philip Tetlock tracked thousands of expert predictions and found well-credentialed experts did barely better than chance. "20 years of experience" can mean 20 years of skill, or one year repeated 20 times
Credentials Degrees and licences are a shortcut They show what someone has learnt, not whether they are good at the task on a given day. Is a surgeon's board certification better evidence than their actual surgical skill?
Peer recognition Awards, citations, reputation matter because often only insiders can judge insider work Fields fall in love with fashionable ideas and reward fitting in over being right

All four signals share something most of us cannot check on our own, so we must keep asking what lies underneath the signals we encounter.

Types of expertise

Type Meaning Example
Contributory Actually does the work (Collins and Evans) Surgeon; physicist running an experiment
Interactional Cannot or does not do the work, but has absorbed the field's language well enough for an expert conversation A great science journalist
Explicit Knowledge that can be written down A recipe
Tacit Cannot be fully written down; travels through practice and apprenticeship, so nobody becomes an expert from a manual alone. May carry no credential A nurse who senses a patient worsening; a carpenter who sees something wrong in the wood
Experiential Informal knowledge from lived experience; does not replace formal knowledge but improves decisions A 20-year migraine sufferer
Meta-expertise The skill of judging experts when you are not one yourself; what we need most of the time Evaluating sources

Meta-expertise depends on your surroundings: you can be good at evaluating sources and still be poorly served if someone else chose your sources or taught you to distrust good evidence (Manan).

Whenever someone says "the experts say...", ask: Which experts? Expert in what exactly? How did you come to trust them?

Earning trust as an expert

Admitting limits can make an expert more believable. "I am confident about this part, less about that part" or "that is outside my area" earns more trust than claiming to know everything and making things up when pressed. Overconfidence works, but not always.

Habits of a trustworthy expert:

  1. Show your reasoning.
  2. Stay in your lane.
  3. Disclose your incentives.
  4. Make mistakes clear when you make them.

Critiquing experts

Questioning experts is legitimate and necessary. Experts can be:

  • Overconfident, and frame problems through their own narrow tools.
  • Subject to vested interests and funding pressures.
  • Groupthinkers, influenced by what their own group says.

Finding these blind spots is the critical thinker's job.

Key takeaways

  • Maya is in an epistemic bubble (poppable); Manan is in an echo chamber (pre-discredited outsiders).
  • Expertise is seen through networks of trust, not experts alone.
  • Knowledge, track record, credentials and peer recognition are signals that each can mislead.
  • Track records help only when feedback is quick and clear (Tetlock).
  • Know the types of expertise, especially tacit and meta-expertise.
  • Ask: which experts, expert in what, why do I trust them?

How Value Is Created: The Social Construction of Value

The Joshua Bell experiment

In January 2007 a man in jeans and a baseball cap played violin for about 45 minutes in a Washington, D.C. metro station. He was Joshua Bell, one of the world's best violinists, playing an instrument worth millions. Days earlier he had sold out a concert hall where good seats cost well over $100. In the station almost nobody stopped; he collected a little over $30.

Same man, same music, same violin. The difference was not the sound but the setting: the hall, the ticket, the reputation and the crowd that signal "this is worth attention and money."

Thesis: value is not purely intrinsic, like the weight or colour of an object. Much of it is built by institutions, markets, brands and reputations. This does not make it fake; it means it is made.

Intrinsic vs. extrinsic value

Meaning Example
Extrinsic Value as a means to something else A hammer matters only when used on a nail
Intrinsic Value for its own sake Happiness, friendship, a person's life

Two cases show how much "value" is added from outside:

  • The diamond. In the 1940s De Beers and its ad agency ran a campaign around the tagline "A diamond is forever," reshaping the engagement ring market. The carbon was always there; the meaning and much of the price were added.
  • The Mona Lisa vs. a skilled copy. Side by side, an original and an identical-looking copy may be indistinguishable to the eye. Yet the original is prized for its history, authenticity and place in a story, all of which are relational.

Much of a thing's value is a fact about its relationship with the viewer, listener or receiver. It is socially constructed, but not arbitrary or fake: money is socially constructed yet buys real things. The useful questions are who builds value, how, and whether we notice it is built.

How institutions create value

  • Museums confer value, not just display it. A glass case, a light and a label tell visitors "pay attention, this matters."
  • Symbolic or cultural capital (Pierre Bourdieu): cultural fields run on a currency that is not money. Critics and curators decide what counts as good taste.
  • Rankings do not merely measure schools; they change how people work within the institutions they rank.
  • Certifications (organic labels, Michelin stars, fair trade marks, academic accreditation) are good solutions because we cannot judge quality ourselves. The label stands in for knowledge we lack, which is useful, but it also draws a line: we may trust the label more than our own judgment.

Value, trust and fragility

  • A story and an endorsement can build a lot of value, but not forever. If the claim the value rests on is false, the value falls apart. Spotting this is the critical thinker's job.
  • Reputation and legitimacy are the social forms of the credibility discussed for experts. They let strangers rely on each other without checking everything. If an "organic" label has held up when you ate the product before, you can trust it again without re-checking.

Museums, critics, experts, rankings, prizes, brands and reputations are useful, but each is also a lens that shapes what and how we see.

Key takeaways

  • Bell's sound did not change; the setting, ticket, reputation and crowd did.
  • Extrinsic value is a means to something else; intrinsic value is for its own sake.
  • Diamonds (De Beers) and the Mona Lisa show value is relational and added.
  • Institutions (museums, rankings, certifications) confer value; Bourdieu calls this symbolic or cultural capital.
  • Labels substitute for knowledge we lack, but can displace our own judgment.
  • Value built on a false claim collapses.

Expertise in Practice: A Professional Researcher's View

Conversation with Charan Tadepalli, who has over 16 years in consumer research and now works in modern UX research.

Expertise in Practice: A Professional Researcher's View

The UX researcher's role

Tech organisations (Google, Meta, Microsoft, ServiceNow, Oracle) stand on three pillars:

Pillar Owns
Engineering The backbone of the product (code)
Product Strategy and market position
Design Look and feel of the interaction

Research is "the blood that flows across" all three: each needs customer feedback. A senior researcher joins conversations on product strategy and organisational vision and, most importantly, acts as an unbiased, neutral voice in the room. Day-to-day work includes recruiting and identifying respondents, choosing methodology and markets, logistics, moderating conversations, then analysing and synthesising data and tying it back to what the business needs, while reporting accurate findings even when they are not what the business wants to hear.

  • Each function owns something and can become territorial about it (code, strategy, UI/UX), including researchers about their own documents.
  • Territoriality mostly appears when feedback is unstructured and not "standing on strong legs." These teams are also open, since their shared agenda is business growth.

Research rigor as the source of credibility

  • Analogy: people accept a medical research paper because they trust the rigor behind it (testing, control and test groups), while still sometimes checking for industry lobbying.
  • A researcher's strongest leg is rigor: "You're not trusting me; this is what 100 people have said." He ties this to a sample large enough for roughly a 95% confidence level, shrinking the margin of error. (He gives this as a general statistics point; no calculation is shown.)
  • Data is not infallible. Quoted Henry Ford: people asked what they want would say "a flying horse," and could not have imagined the automobile. Steve Jobs famously did not research the iPhone with consumers.
  • Where no benchmarks exist (for example a brand-new AI feature, unlike a soap with sales history), look at the closest existing behaviour. Example: before AI summarises HR, IT and admin tickets, study how humans did that summarisation, its difficulty, need gaps and pain points, and feed that into the solution.
  • When there is disagreement: remember you are a partner; rely on data, numbers and consumer verbatims; if stakeholders still proceed, become a voice of warning naming what could go wrong and immediate actions to take.

Structure and simplification

  • Researchers are trained: "don't complicate"; "Tell me what's the story?"; a senior colleague's rule was "no death by data; inspire me with insight." Explain as you would to a child, breaking down the logic, without treating stakeholders as children.
  • Business owners want a few clear points (for example the three things to do). The simplification chain:
  1. 1Data
  2. 2Patterns
  3. 3Dominant patterns
  4. 4Human behaviour behind them
  5. 5Business implication
  6. 6Recommended action
  • Organisational conversation is like interpersonal conversation but with formality and decorum, which is why boards and markers are used to visually map thoughts.
  • Three languages starting with C: engineering speaks code, product speaks company, research/design speak customer. Each struggles to understand the others, so findings must be translated both ways.
  • Frameworks must be applied fluidly on the ground, with other humans in the room.

Credentials versus credibility

  • Early in his career (around age 22 or 23), on a newspaper-positioning project that reached his agency via a Big Four consulting firm, a consultant from a top institute treated him as "just a data vendor" because his MBA lacked a pedigree brand. His manager's advice: be the loudest voice in the room, or build credibility over years of consistent work.
  • Later, with 16 years of experience and work with names such as Microsoft, Flipkart, Kantar and Nielsen, credentials ("I'm from Cornell") matter less, but even then a half-baked point of view can be challenged by anyone, even front-desk staff. Hence rigor and groundwork are always required.
  • Practices: a three-line elevator pitch even for a 100-slide report; tailor to engineer, product manager and designer (product, industry numbers, user experience); identify three key takeaways; use them in emails and invitations. The host calls this empathy: attempting to know what the other person wants.
  • The core of research is observation and listening, which is empathy; a good researcher speaks little. Thinking beyond your credentials and being aware of your own biases are the link back to critical thinking.

AI, trust and value

  • His work has shifted from AI capabilities to value in AI: what metrics show value (man-hours saved, labour cost saved, more tasks per employee, dollars saved, new ventures)? Answers boil down to trust.
  • Using the transactional analysis model (parent, adult, child): people assume AI is intelligent because of its name, but it is only as good as its data. An ideal exchange is adult-adult; often it becomes parent-child, with the human as parent and AI as an uninformed child, causing frustration, trust issues and broken workflows.
  • A study he attributes to BCG (he is not certain) reported that only about 5% of organisations that invested in AI see five times more returns, because they cleaned their data and rebuilt workflows. Typical organisations hold information scattered across email, Slack, Teams and phone calls, with no clear provenance or approval record.
  • AI works on procedural intelligence (algorithm, process); humans run on jugaad. Their sparring creates trust problems.
  • Empathy for AI is neither expected nor wanted: it is a tool, like a laptop or calculator. Trust, yes; empathy, no.
  • Forward-deployed engineers from OpenAI and Anthropic (like Cognizant and Infosys consultants) are placed in companies to clean data and prepare them for AI.

Lesson: never trust a finding at face value

The Hindu case. Times of India ran a Tamil, Chennai-set ad mocking The Hindu as boring (showing a man asleep). Research for The Hindu found it read as intellectual: used for exam preparation (GRE, TOEFL, IELTS), loyal among a small audience valuing its thought leadership, but the larger spending population did not think of it that way. The instinctive response is to make it snazzier. Instead, working with Ogilvy & Mather, the campaign "Stay ahead of times" reframed the perceived weakness as a perception problem: the problem is not the paper but how people think, delivered with humour and wit. (A quiz-style ad with humorous wrong answers ended with "What newspaper do you read?") The paper also made real changes (colour pages, wider topics).

The Rapido case (2018). Rapido, newly funded and expanding, wanted awareness. The working frame was positioning against Uber and Ola. He questioned it: Uber/Ola users would never ride pillion with a shared helmet in dust and pollution, so the real target is the person toiling in the RTC bus. He built "commute currencies" (pride, convenience, comfort) to set context. The resulting campaign (with Allu Arjun and Ranveer Singh, "starts at 19 rupees") worked well.

The fragrance case. A large Indian conglomerate wanted to enter fragrance. After about 200 interviews across the country and an award-winning study, the client asked only for a choice from the standard fragrance wheel, copying what Gucci, Calvin Klein and Yardley do, because of a fixed launch date and an existing supply chain. The researchers feared a "me too" brand. Compromise: a short-, mid- and long-term plan: use the existing supply chain for the immediate launch on the option closest to the brand, change packaging and communication to avoid looking like a copy, and build new supply chains over time. Their job expanded to include supply-chain thinking because that was the client's context.

Framing recommendations. Numbers are "only 30%" of the story: data gives accuracy, the narrative influences decisions. Establish context first (who we met, what we observed, why), so the recommendation is expected by the time it arrives. Data is the foundation, not the whole building.

Closing habit: "Stay frustrated, stay intrigued" (a play on "Stay hungry, stay foolish"): without curiosity you are not growing.

Key takeaways

  • Research acts as a neutral voice among engineering, product and design, and earns credibility through rigor.
  • Structure and simplification (data, pattern, behaviour, implication, action) make findings usable.
  • Credibility is built by consistent work and groundwork, not credentials alone.
  • Trust in AI depends on data quality and workflows; treat AI as a tool, not an empathic partner.
  • Never accept findings at face value; reframe problems and be open to the question changing.
  • Numbers are a strong foundation, but framing and context drive decisions.
Revise before your examAll 3 exam tips from this module in one list
  1. Expertise is not any single one of the four signals. Each is evidence and each can mislead; the track record is strongest but only when feedback is clear and quick.
  2. "Socially constructed" does not mean "fake" or "arbitrary." The lesson of the metro station is that value is made, by people and institutions with their own expertise and vested interests.
  3. The question can change as you learn, but the answers must remain truthful. Be open to the question itself changing.

Test yourself on this module

Questions from these notes, one at a time, with feedback. Your weak topics are saved.

Set it up free Takes about a minute. Free plans work.