Taste

Vital signs

Why it's on the table

On the table, Taste (Ts) is seat 57 of 58, in the Human Elements family. It is a stable element — the category has settled, choosing is cheap, and switching is rare. Pick a holder and move on; this is not where your decision budget should go. It is marked essential: a company that leaves this seat empty is running with a gap that compounds. The price of entry is zero, which makes trying it a decision that needs no meeting.

Taste: the top 5 — v2026.Q3

Edition v2026.Q3 · ranking, pricing and status verified 2026-08-06.

1. 1Deliberate study of a named canonyou + Rams · Tufte · Zinsser · Ogilvy · Graham

   Free (web + library) · ~$150–400 to own the canon in print · Tufte's course $240 ($160 student) incl. five books · 2–5 h/week, indefinitely

   Best for The founder who can tell that something is off but can't say why — the gap between reaction and articulation, which is what a canon closes.

   Taste is discrimination before it is production: Rams' ten principles (formulated late 1970s, still published verbatim by Vitsœ) and Graham's fourteen properties of good design (Feb 2002) are the two shortest documents that will change what you can see, and both are free. The move is active, not passive — take a page of Zinsser, a Tufte chart, an Ogilvy headline and reconstruct why each decision was made, in writing. Do that and you acquire vocabulary; vocabulary is what converts a vague 'this feels wrong' into a specific instruction an agent can execute, which is precisely the bottleneck when production is free.

   Watch The evidence for structured practice producing expertise is weak exactly where founders live: deliberate practice explains 26% of performance variance in games, 21% in music, 18% in sports, 4% in education — and less than 1% in professions (Macnamara, Hambrick & Oswald, Psychological Science, 2014). The failure mode is connoisseurship: people who can critique anything and ship nothing. Dan Pink's counter-argument (Inc, Jun 17 2026) is that 'creation beats consumption' and taste comes from making, not studying. Both are right; study without rank 2 is a hobby.

   Rams' ten principles, formulated late 1970s, still the most-cited design canon (Vitsœ, publisher of record) [src] · Paul Graham, 'Taste for Makers' (Feb 2002): 14 properties of good design shared across disciplines [src] · Honest counterweight: deliberate practice explains <1% of performance variance in professions, 4% in education (Macnamara et al., Psych Science, 2014) [src]
   Site →Docs →

2. 2A structured critique looprun it yourself (Pixar Braintrust model) · or buy it: Shift Nudge critique vault · Maven cohorts · writing workshops

   Free if you run it — 3–6 peers, 60–90 min weekly, work shown before opinions · Shift Nudge $1,997/yr (bi-weekly live critique + 1,000+ recorded reviews) · Maven cohorts $1,100–2,700

   Best for Founders already shipping a lot whose work has stopped getting better — the classic plateau, which is always a feedback problem, never an effort problem.

   This is the half of 'deliberate practice' that people skip, and it is the half that carries the causal weight: Kluger & DeNisi's meta-analysis (607 effect sizes, 131 papers, 12,652 participants) found feedback improves performance on average at d = 0.41. Pixar's Braintrust codified the structure that makes it work — candid, peer-level, aimed at the work, with no authority to mandate fixes — and the commercial versions copy it: Shift Nudge's core asset is a vault of 1,000+ recorded design critiques, not its video lessons. A critique loop is also the fastest way to discover that your taste is confidently wrong, which is the discovery no amount of reading produces.

   Watch The same meta-analysis found over 38% of feedback interventions decreased performance — feedback aimed at the person rather than the task reliably backfires. Founder-specific failure: a critique group of your own employees is not a critique group, because nobody in it is free to say the product is bad. Paid critique buys you strangers, which is most of what you're paying for.

   Feedback interventions: mean d = 0.41, but >38% of effects were negative (Kluger & DeNisi, Psychological Bulletin, 1996; 607 effect sizes, 131 papers, 12,652 participants) [src] · Shift Nudge: 4,000+ designers, 1,000+ recorded critiques, bi-weekly live critique calls for 12 months (Aug 2026) [src] · Doshi's Maven product-sense cohort: 1,000+ students, median 12+ years' experience — a paid critique room, priced accordingly ($2,700) [src]
   Site →Docs →

3. 3A curation dietyou (input hygiene) — Kleon's 'Steal Like an Artist' is the canonical brief

   Free — costs attention, not money · ~30–60 min/day of chosen input, plus the unsubscribes and mutes that make room for it

   Best for Work that is competent and generic — the specific failure of a founder whose entire input stream is an algorithmic feed optimized for engagement rather than quality.

   The argument is simple and, in 2026, newly urgent: your output distribution is bounded by your input distribution, and the input distribution has been flooded. Graphite's May 2026 analysis puts roughly half of all articles published online as AI-generated; 'slop' was Merriam-Webster's 2025 Word of the Year, defined as low-quality digital content produced in quantity by AI. If your reference set is the median of that, so is your product. The practice is deletion first — cut the feeds — and replacement second: a small number of deliberately chosen sources, at least some from outside your industry, because cross-domain reference is where non-obvious product decisions come from.

   Watch This is the thinnest-evidenced pick on the page. No controlled study links consumption quality to output quality; the entire case is analogy and anecdote, and we are grading it D on purpose. It is also the practice most easily faked — reading design newsletters is consumption cosplaying as practice, which is exactly the 'tastewashing' NoGood documented in 2026: performing taste is cheaper than developing it. Dan Pink's 'creation beats consumption' (Inc, Jun 17 2026) is a direct rebuttal and should be taken seriously.

   'Slop' — Merriam-Webster's 2025 Word of the Year: 'digital content of low quality that is produced usually in quantity by means of artificial intelligence' [src] · Roughly half of all articles published online are now AI-generated (Graphite research, May 2026) [src] · 42% of consumers say low-quality or 'uncanny' AI-created advertising would negatively affect their opinion of the brand; 56% can't reliably identify AI content (DoubleVerify, Jul 29 2026; 22,000 consumers, 22 markets) [src]
   Site →Docs →

4. 4Saying no · editing downyou (the core exercise) — Rams' tenth principle, 'less, but better'

   Free — but the real price is revenue you decline, features you kill, and customers you disappoint. Time cost: near zero. Effect: immediate.

   Best for The AI-era default failure: a founder shipping a great deal, fast, with no discernible point of view — because every generated option looked plausible and none were refused.

   Every other practice here is additive and slow; this one is subtractive and instant, which makes it the highest-leverage thing on the page. Rams' tenth principle — 'less, but better, because it concentrates on the essential aspects' — and Graham's first property, simplicity, are the same instruction from two disciplines. Dan Pink's formulation (Inc, Jun 17 2026) is the operational one: taste is 'saying no. No, that's not good enough. No, that's not for me.' When generation is free, the only scarce act left is refusal, and it is the one act you cannot delegate to a model that was trained to be helpful.

   Watch The evidence is case-study and aphorism, not study — we grade it C, and the widely-repeated Jobs 'cut 350 products to 10' figure is folklore with no primary source we could verify. The failure mode is symmetric and worse than the disease: founders who say no to distribution, to unglamorous customers, and to shipping, then call it taste. Saying no is only taste if you can articulate the principle it serves; otherwise it is preference with better branding.

   Rams, principle 10: 'Good design is as little design as possible — less, but better' (formulated late 1970s; Vitsœ, publisher of record) [src] · Dan Pink (Inc, Jun 17 2026): taste requires the courage to decline — 'No, that's not good enough. No, that's not for me' [src] · Paul Graham (quoted Feb 2026): 'When anyone can make anything, the big differentiator is what you choose to make' [src]
   Site →Docs →

5. 5Formal craft trainingShreyas Doshi (Maven) · Shift Nudge · Reforge · NN/g · IDEO U · Section

   Advanced Product Taste $1,100 (6 h, Oct 17 2026) · World-class Product Sense $2,700 (8 days, Sep 26–Oct 4 2026) · Shift Nudge $1,997/yr · Reforge $1,995/yr · NN/g UX Certification ~$3,110 (5 courses + exams), Master ~$9,330 · IDEO U $199–899/course, $999 all-access · Section Premium $62.50/mo billed annually · Refactoring UI $99–149 · Ship 30 for 30 $350

   Best for Founders who need a syllabus and a deadline rather than discipline — and technical founders with no visual training at all, for whom tactics beat theory.

   Paid programs buy three real things: reps, strangers who will criticize you, and a vocabulary you'd otherwise assemble over years. The price ladder is honest about what it sells — Refactoring UI at $99–149 (30,000+ copies) is pure tactics for developers who ship ugly interfaces; Shift Nudge at $1,997/yr is craft plus a standing critique room; Doshi's $1,100 one-day workshop and $2,700 cohort are framework-and-peers for people already senior enough to use them. What none of them sell is taste itself, and the good ones say so.

   Watch Not one provider in this category publishes a controlled outcome study; every claim is testimonial. The cohort model's economics are also genuinely fragile — Write of Passage, the most celebrated writing cohort of its era (2,000+ students, 72 countries), shut down after its Oct–Nov 2024 final cohort because, per Perell, the business didn't work. Learn UI Design's individual enrollment is closed as of Aug 2026, waitlist only. Treat any 2026 recommendation of a cohort course as perishable, and prefer the ones with a permanent artifact (book, vault, recordings) over the ones that are only a calendar event.

   Advanced Product Taste (Shreyas Doshi, Maven): $1,100, 6 hours live, Oct 17 2026 — 'Product Taste in 1 day: Practical, Advanced, AI-pilled' [src] · NN/g UX Certification: ~$3,110 for 5 full-day courses + 5 exams (30 h min); Master ~$9,330 for 15 courses [src] · Graveyard: Write of Passage shut down after its final cohort Oct 7–Nov 13 2024 — 2,000+ students, 72 countries, killed by cohort economics [src]
   Site →Docs →

Taste: the top 8 compared

Edition v2026.Q3 · ranking, pricing and status verified 2026-08-06.

How to choose your taste

If you have zero budget and one hour a day

Deliberate study plus a weekly critique group with 3–6 peers. Free, and the only pair here with a legible mechanism — Rams and Graham take an evening to read; the critique loop is what converts reading into judgment.

If you're shipping constantly and the work has stopped improving

A critique loop, not more input. Structure it against the task, never the person — Kluger & DeNisi found >38% of feedback interventions made performance worse, and person-directed feedback is the reliable way to land in that 38%.

If you're a technical founder shipping interfaces you know are mediocre

Refactoring UI ($99–149) first — tactics transfer faster than theory when you have no visual training — then Shift Nudge ($1,997/yr) if it's still mediocre in three months. Skip the $3,110 certification; you need craft, not a credential.

If you're choosing between one $2,700 cohort and twelve months of a free critique group

Take the critique group. Cohorts front-load motivation and end; taste is built by the reps after the cohort ends — which is roughly the reason Write of Passage's economics failed despite 2,000+ satisfied alumni.

If someone is selling you taste as a one-day workshop, or your investors are telling you taste is the moat

Buy the framework, not the promise, and discount the rhetoric. No provider in this category has published an outcome study, and SignalFire's 2026 data shows design hiring down 48% at tech majors — the people loudest about taste are not the ones funding headcount for it.

Taste: the whole field

24 more tools tracked in this category, including 3 dead, renamed, or sunsetting — a reference that hides the graveyard isn't one. Verified 2026-08-06.

Taste: the category in numbers

Edition v2026.Q3 · ranking, pricing and status verified 2026-08-06.

• 'Slop' was Merriam-Webster's 2025 Word of the Year — 'digital content of low quality that is produced usually in quantity by means of artificial intelligence.' The dictionary named the problem this element exists to solve [src]
• DoubleVerify global study (Jul 29 2026; 22,000 consumers across 22 markets, 2,020 marketers across 21): 42% say low-quality or 'uncanny' AI-created advertising would negatively affect their opinion of the brand; 56% cannot consistently identify AI-generated content; 53% of marketers worry about ads running beside poor-quality AI content [src]
• Taste entered the executive vocabulary inside two weeks, Feb 2026: Greg Brockman — 'Taste is a new core skill'; Sam Altman — 'the best research teams are built through context, taste and a real feel for where the field is headed next'; Paul Graham — 'when anyone can make anything, the big differentiator is what you choose to make'; Cloudflare CTO Dane Knecht — 'building is easy now, knowing what to build is the hard part' [src]
• Counter-signal — SignalFire State of Tech Talent 2026: design hiring down 48% at tech majors and 22% at startups; marketing down 36%/18%; front-end engineering down ~25%, the steepest engineering drop, as backend generalists generate UIs. New-grad hiring down ~65% at majors and ~76% at early-stage startups vs 2019 [src]
• Graphite research (May 2026): roughly half of all articles published online are now AI-generated — the supply-side reason discrimination is repricing upward [src]
• The strongest quantitative case for design-as-differentiator is still eight years old: McKinsey Design Index (Oct 25 2018; 300 public companies, 5 years, 2M+ financial data points, 100,000+ design actions) found top-quartile design performers grew revenue 32 percentage points faster and TRS 56 points faster than industry peers [src]

Taste: method & sources

Ranking charter: mechanism first, transferability second, cost last — which is why two free practices outrank every paid program, and why the paid rank sits at 5. No affiliate consideration; picks are editorial. Evidence honesty, stated plainly: this is the weakest-evidenced element on the table and we are not going to pretend otherwise. There is no randomized trial of 'taste training' anywhere in the literature. The two studies we lean on are both borrowed from adjacent domains and both cut against the popular story — Macnamara, Hambrick & Oswald (Psych Science, 2014) found deliberate practice explains 26% of variance in games, 21% in music, 18% in sports, 4% in education and under 1% in professions; Kluger & DeNisi (Psych Bulletin, 1996) found feedback helps on average (d = 0.41) but that over 38% of interventions made performance worse. Ericsson disputed the Macnamara reading on definitional grounds (Frontiers in Psychology, 2019, arguing the original definition of deliberate practice was not applied) and we cite Macnamara anyway, because the looser definition is the one founders actually practice. Grade the category C overall, D for the curation-diet pick. Conflicts resolved, not hidden. (1) Study vs. making: Dan Pink's 'creation beats consumption' (Inc, Jun 17 2026) directly contradicts the curation-diet thesis; we kept both and ranked study-plus-critique above diet accordingly. (2) Rhetoric vs. hiring: the Feb 2026 taste discourse says taste is the scarce input, while SignalFire's 2026 data shows design hiring down 48% at tech majors — we report both rather than picking the flattering one, because the honest synthesis is that taste is being reassigned to founders and senior operators, not staffed. (3) The 'tastewashing' critique (NoGood, 2026) — that companies invoke aesthetic conviction while practising market discernment, and that performing taste is cheaper than developing it — is the strongest attack on this whole element and we consider it largely correct as a description of the discourse, and irrelevant as an argument against the practices. Unverified or soft: book list prices are publisher/list figures, not checkout-verified on 2026-08-06 (HarperCollins blocked our fetch with a 403; Penguin's Creativity, Inc. page returned 404, so we linked the encyclopedia entry). Tufte's $240/$160 pricing sits on a page last dated Aug 2024 — plausible but not freshly confirmed. Graphite's 'half of all content' figure comes from a research index page without exposed methodology; treat as directional. The Jobs '350 products down to 10' figure is folklore with no primary source and is deliberately excluded from evidence. Maven course prices are per-cohort and change between sessions. Adjacent elements: design tooling and production (Figma, Canva) → De; prompt-and-image craft → Im; credentialing your taste for a market → Sk; the peer group that makes a critique loop possible → Co; senior judgment you rent instead of build → Fx and Mn. This element covers only the founder's own faculty of judgment and the practices that develop it. Ranking criteria: verified commercial traction, independent satisfaction surveys, agent benchmarks, and founder-fit (price floor, lock-in, surfaces). Editorial, never paid — the charter. Machine-readable twin: ts.json.

Our take

Every other element got cheaper. Taste got scarcer. It's the last durable moat a founder personally owns.

Combines with