# Design Taste — Complete Notes, Drills & Implementation Plan **Compiled for Johan · 14 November 2026** **From five Medium articles, read in full.** Nothing paraphrased away: every claim, number, example and named source is kept. **Also available:** [Word version (printable)](Design-Taste-Complete-Notes.docx) · [Markdown source](notes-master.md) · [Full unedited article text](medium-design-taste-full-articles.md) Every line below is tagged so you can act on it rather than just read it: - **REMEMBER** — a principle worth committing to memory (the *why*). - **PRACTICE** — a drill that builds the skill (the *rep*). - **IMPLEMENT** — a change to how you actually work, review or brief (the *system*). - ☐ — a checkbox: something concrete to tick off. --- # Part 1 — Article-by-article notes ## 1. How to Develop Design Taste in the AI Era: A 4-Week Plan for UX Designers **Nishkarsh Gupta · ILLUMINATION · Oct 2026 · ~1,672 words** [medium.com/illumination/design-taste-new-skill-how-to-build-it-a46d424f3173](https://medium.com/illumination/design-taste-new-skill-how-to-build-it-a46d424f3173) **The author's stake:** he works mostly in B2B, so his examples and conclusions lean toward tables, forms and empty states rather than hero images. ### 1.1 The definition of taste — and why explanation is the whole point - **REMEMBER:** Design taste is **trained judgment**: the ability to tell good work from *almost* good work, and to **explain why in plain words**. It is not a gift you are born with. - **REMEMBER:** You build it by doing three things: looking at a lot of work, thinking about it, and making things yourself. - **REMEMBER:** Taste is **the skill of seeing that one screen works and another does not, then saying what is different** — the spacing, the order, the tone, the one thing that is missing. - **REMEMBER:** Taste is not "I like it." Taste is *"I like it because the label sits close to the field, the contrast is clear, and nothing else fights for attention."* - **REMEMBER:** **If you cannot explain your choice, you cannot repeat it, and you cannot teach a tool to do it either.** Explanation is the transfer mechanism. - ☐ Write a one-sentence "because…" for the last three screens you admired. If you can't, that's your first drill, not a knowledge gap. ### 1.2 The Ira Glass trap - **REMEMBER:** Public radio host **Ira Glass** described the trap most of us fall into: **people start creative work because they have good taste, but their early work is not yet as good as their eye.** That gap is normal. - **REMEMBER:** Glass's advice was short: **"Do a huge volume of work."** The gap closes by making things, not by waiting. - **IMPLEMENT:** When work disappoints you, name it as the gap (eye ahead of hands), not as evidence you lack taste. Volume, not reflection alone, closes it. ### 1.3 What taste is *not* - **REMEMBER:** It is **not about style trends**. Not expensive gear. Not the same as being trendy. Not the same as being minimal. - **REMEMBER:** **A crowded dashboard can have great taste if every row earns its place.** (Contrast this with the "minimal = tasteful" reflex.) - **REMEMBER:** In B2B work, taste shows up **in tables, forms and empty states far more than in hero images**. - ☐ Audit one of your own dense screens for "does every row earn its place?" rather than for tidiness. ### 1.4 Why taste matters *more* when AI can generate screens - **REMEMBER:** A decent-looking screen is now **cheap**, so what's left is **the choosing**. When everyone can generate ten options in a minute, the person who picks the right one — and knows what to change — has the edge. - **REMEMBER — NN/g, State of UX 2026** (January 2026, by **Kate Moran, Raluca Budiu, Sarah Gibbons** and the NN/g team): design systems, patterns and components have made interfaces more standard, and as AI design tools improve, **that standardisation gets stronger and almost anyone will be able to make a decent-looking UI**. Their blunt line: **if you only put together components from a design system, AI can already replace you.** - **REMEMBER:** NN/g then name what is hard to automate — **curated taste, research-informed understanding of context, critical thinking, careful judgment**. Taste is *one of four*: it sits **beside** research and thinking, not above them. - **REMEMBER:** **Good taste with no understanding of users only gives you a pretty wrong answer.** - **REMEMBER — UC Berkeley School of Information, 2026 final project:** watched **13 UX designers** use prompt-based tools like **Lovable** to build interfaces and found **six kinds of breakdown**. Two matter most: - **The vocabulary gap** — designers could not find the right words for what they wanted. - **The convergence trap** — the tool drifts toward generic, polished results. - **REMEMBER:** In short: **the tool gives you the average, and only your eye can push it away from average.** - ☐ In your next AI-assisted exploration, log every time you wanted something specific but wrote something vague — that's your vocabulary gap, measured. ### 1.5 The five habits (the author's real thesis) - **REMEMBER:** You build taste with **five small habits done often**: (1) collect references on purpose, (2) write down why you like them, (3) copy to learn, (4) critique every day, (5) look outside software. - **REMEMBER:** **None of them needs a course. All of them need repetition.** ### 1.6 Habit 1 — Collect references on purpose - **REMEMBER:** Start with a library, but treat it like a **study desk, not a wall of pretty pictures**. - **REMEMBER — named sources:** - **Mobbin** — holds **more than 620,000 screens and over 300,000 flows** from actual apps; lets you save designs into collections and **add notes when you save**. - **Wall of Portfolios** — a curated set of UX, product and interactive design portfolios, filterable by **role, experience and theme** (e.g. minimalist, dark). - **REMEMBER:** The rule that helps most: **save fewer things, and name the folder after a question, not an app.** *"How do good apps handle empty states?"* teaches more than *"Stripe stuff."* - **IMPLEMENT:** Rename your three largest reference collections into question form this week. - ☐ Convert one collection to question-naming and add notes to the first 10 items. ### 1.7 Habit 2 — Write down why you like it (the habit that converts feeling into skill) - **REMEMBER:** Under every saved screen, write **three short lines**: ``` What I see: ____________ Why it works: ____________ What I would steal: ____________ ``` - **REMEMBER:** This **fixes the vocabulary gap** from the Berkeley study. **The words you build here are the words you will later use to direct a person, a teammate or an AI tool.** - **REMEMBER:** *"Make it cleaner"* gets you the average. *"Give the heading more space above and cut the border on the cards"* gets you closer to what you meant. - **IMPLEMENT:** Adopt the three-line note as a hard requirement on every save, in whatever tool you use. - ☐ Write three-line notes for 5 references today and reread them tomorrow — do they read as instructions someone else could follow? ### 1.8 Habit 3 — Copy to learn - **REMEMBER:** Pick **one screen you admire and rebuild it by hand in Figma**. **Do not trace it with a plugin.** - **REMEMBER:** Measure the spacing, guess the type sizes, match the colors, and look at **how the layout is built**. You will find out **how much you missed when you only looked**. - **REMEMBER:** The point is **not to publish the copy** — the point is to **learn the decisions behind it**. - **REMEMBER:** When you finish, **change one thing on purpose** (the type scale, the density) and ask whether it got better or worse. - **REMEMBER:** The analogy: it is like **learning other people's songs before you write your own**. - **REMEMBER:** One extra step makes copy stronger: **before you start, write down three guesses about why the screen looks good. After you finish, check which guesses were wrong. The wrong ones are where you learned something.** - **PRACTICE:** 90-minute handrebuild, no plugins, three guesses written first — see the drill sheet in Part 6. - ☐ Schedule the first handrebuild and write the three guesses before you open Figma. ### 1.9 Habit 4 — Critique every day - **REMEMBER:** **Ten minutes is enough.** Open one screen and answer **three questions**: what works, what does not, and what would I change first. **Say it out loud or write it down.** - **REMEMBER:** **Critique AI output too.** Generate a screen in your favorite tool and look for where it is generic: **Is every section the same weight? Is the copy vague? Does the hierarchy tell you where to look first?** - **REMEMBER:** This is where taste pays off in daily work: **you can now edit the machine instead of accepting it.** - **REMEMBER:** A tip for the daily critique: **keep one running note, and add a rule to it whenever you catch yourself saying the same thing twice.** After a month you have **a personal style guide written in your own words.** - **IMPLEMENT:** Half your critiques on human work, half on AI output (this split recurs in the 4-week plan). - ☐ Start the running rules note; add the first rule the second time you catch a repeat. ### 1.10 Habit 5 — Look outside software - **REMEMBER:** **If you only study apps, you will only make things that look like apps.** - **REMEMBER — sources named:** posters, book covers, packaging, street signs, maps, film title cards and buildings. - **REMEMBER:** Take a photo, then **ask the same three questions**. - **REMEMBER:** Outside references give you ideas **the rest of the crowd does not have**. Since AI tools **lean toward the average of what already exists**, a fresh source is one of the few ways to end up with something that **does not look like everyone else's**. - ☐ Photograph 10 non-software references and run the critique questions on each. ### 1.11 The 4-week plan (verbatim structure) - **REMEMBER:** **20 to 30 minutes a day for four weeks**, one job per week. - **Week 1 — collect and name.** Save **five screens a day**. Write **the three lines** under each one. By the end you have **about 35 notes and a shared vocabulary**. - **Week 2 — copy.** Rebuild **three screens by hand, one every two or three days**. After each, **write down two rules you found**. - **Week 3 — critique.** **Ten-minute critique every day.** Do **half on human-made work and half on AI-generated screens**. - **Week 4 — look outside and make.** Collect **ten references from outside software**. Then **design one screen using only your written rules**, and **compare it with what an AI tool gives you for the same brief**. - **REMEMBER:** At the end of week four, **read your notes from week one. The difference in how you describe things is the proof that your eye moved.** - ☐ Put the four weeks in the calendar with the daily slot blocked, not just intended. ### 1.12 Can taste be taught? (and the honest limit) - **REMEMBER:** **Yes, in the sense that matters for work.** You cannot fake it in a weekend, but **attention, words and volume** make it grow. - **REMEMBER:** The Ira Glass idea holds: **your eye starts ahead of your hands, and steady work closes the gap.** - **REMEMBER:** One honest limit — **taste is not a shortcut around research.** NN/g lists it beside contextual understanding and critical thinking for a reason. **Use taste to judge how something feels and looks, and use users to judge whether it works.** - **REMEMBER — the short version:** *AI makes decent screens cheap. What stays scarce is the person who can tell which one is right and why. Collect on purpose, write your reasons, copy by hand, critique daily and look outside software. Four weeks of small daily effort is enough to notice a change.* - **REMEMBER — the author's closing prompt:** *"Which screen do you admire most right now, and can you say in three lines why it works?"* --- ## 2. Everyone Says Designers Need "Taste". How Do You Actually Develop It? **Life Considered · Muzli — Design Inspiration · Oct 2026 · ~1,345 words** [medium.com/muzli-design-inspiration/everyone-says-designers-need-taste-how-do-you-actually-develop-it-c8c09f0e2605](https://medium.com/muzli-design-inspiration/everyone-says-designers-need-taste-how-do-you-actually-develop-it-c8c09f0e2605) **This is the most practical of the five for a working designer:** it never asks you to collect more references — it asks you to practise on the file you already have open. ### 2.1 The problem it names - **REMEMBER:** You have **saved hundreds of references** and follow designers whose work you admire. **You can recognize good design.** Then you open your own file, adjust the layout, rework the spacing, **and still can't put your finger on why the screen feels off.** - **REMEMBER:** *"Develop your taste"* is easy advice to give and **harder to use** when you're deciding **how much information a screen needs**, **which action should stand out**, or **whether an interaction feels right**. - **REMEMBER:** **Those everyday decisions are the place to practise** — noticing what works, understanding why, and learning when a different approach would serve the user better. ### 2.2 Study more than other interfaces - **REMEMBER:** When your references all come from other apps, **it's easy to keep reaching for the same layouts and visual treatments**. - **REMEMBER — the inputs named:** open a **magazine**, order from a **printed menu**, or **follow signs through an unfamiliar building**. **Pay attention to how the design directs you and where you have to stop and work things out.** - **REMEMBER — worked example (magazine):** a magazine might use **a large headline to invite you into a story, then quieter typography to make a long passage comfortable to read.** - **REMEMBER — worked example (station sign):** a station sign has **a different job** — you need to **find your destination while walking, often with little time to spare.** Notice how **type, spacing and placement serve those situations.** - **REMEMBER:** Bring the question back to your interface: **what does someone need to notice here, and what can stay in the background?** ### 2.3 The capture template for out-of-domain references (4 questions) - **PRACTICE:** When something catches your attention, **save a photo and add a few notes**: - **What did I notice first?** Was it the **size, color, position, or wording**? - **What did the design help me do?** Find a destination, choose an item, or understand where to begin? - **Where did I hesitate?** What was **missing, unclear, or competing for attention**? - **What could I apply to my work?** **Name a specific decision** rather than saving it as "inspiration." - **REMEMBER — the model answer:** *"The station repeats the platform number at each turn. In a long setup flow, I could keep the user's chosen plan visible so they don't have to go back to check it."* - **IMPLEMENT:** This replaces "inspiration" saving. Any saved item without a named specific decision is decoration. ### 2.4 Ask what isn't working — and why (how to request feedback) - **REMEMBER:** *"Looks good"* feels reassuring but **gives you little to learn from**. - **REMEMBER:** When asking another designer for feedback, **explain who the screen is for, what users need to do, and which decision you're unsure about.** - **REMEMBER — the model request:** *"Support agents use this screen to find cases that need attention. I'm unsure whether the urgent cases stand out enough."* - **PRACTICE — then ask these four questions:** - **What do you notice first?** Does it match **what matters most for the task**? - **What is difficult to scan or understand?** Ask them to **point to a specific area**. - **What would you change first?** Ask **what the change would improve**. - **What might that change make worse?** **Consider the tradeoff before applying it.** - **REMEMBER — worked example of the tradeoff probe:** if they suggest **adding more space between rows**, ask **why**. It might help distinguish cases, **but it also means fewer fit on screen.** - **REMEMBER:** **Learning how another designer weighs that tradeoff helps you judge similar decisions in your next project.** Their feedback gives you **a direction to investigate**; **testing with users** can help establish **whether it works**. - **IMPLEMENT:** Ban "looks good / looks nice" as a review response in your own critiques; answer with the four questions instead. - ☐ Rewrite your next review request with the who/what/which decision preamble. ### 2.5 Practice on a problem you can finish (the 5-step loop) - **REMEMBER:** Choose **one part of a current project that still bothers you**: a **crowded setting panel**, a **form with unclear grouping**, or a **table where important details disappear among everything else**. - **PRACTICE — give the practice a clear purpose:** - **Name the problem.** *"The account settings and notification preferences look like one long list."* - **Choose what to work on.** **Start with grouping and spacing before changing colors, icons and typography too.** - **Use realistic content.** Include **long labels and descriptions that wrap onto several lines.** - **Explain the change.** *"I separated the groups so people can see which settings belong together."* - **Check what you sacrificed.** Clearer separation **might make the page longer.** Decide whether **that tradeoff fits the task**. - **REMEMBER:** **Save a short note about what you changed and what still needs to be checked.** Over time **those notes give you a record of your judgment developing**: the problems you notice, the choices you make, and the assumptions you reconsider. - ☐ Keep a `judgment-log.md` in your project: date, problem named, change made, sacrifice accepted, what still needs checking. ### 2.6 Notice how your decisions change (the maturity signals) - **PRACTICE:** **Reopen a design you made a few months ago.** You might immediately want to fix it. **Before you do, write down what you notice.** - **REMEMBER:** *"I don't like it anymore"* tells you little. *"Every card has equal emphasis, so it's hard to tell what needs attention"* gives you **something to work with**. - **REMEMBER — four signs your judgment is maturing:** - **You can name what feels wrong** — you identify a **hierarchy or grouping problem** instead of changing things at random. - **You can explain why a reference fits** — you know **which decision is useful** and **which details don't belong in your product**. - **You recognize the tradeoff** — you can explain **what a change improves and what it makes harder**. - **You revise your judgment when you learn more** — feedback or observed behavior can **change a choice you initially preferred**. - **REMEMBER:** **You'll still make uncertain decisions. Keep track of what you expected, what happened, and what you'd do differently.** - **IMPLEMENT:** Quarterly "reopen and write" session — one screen per quarter, notes before fixes. ### 2.7 Know when your favourite approach doesn't fit - **REMEMBER:** You might prefer **spacious layouts, subtle colors and minimal controls**. Those preferences shape your work — **they can also become habits you stop questioning.** - **REMEMBER — the worked counter-example:** designing a **stock-management screen**, you **hide secondary information to make it cleaner** — but **staff need those details to check each item**, so now **they have to open every row.** - **PRACTICE — four questions before committing to a direction:** - **What task am I making easier?** Name the benefit **beyond "it looks cleaner."** - **What am I making less visible?** Check whether **people need that information to decide or act**. - **Whose preference is driving this?** Distinguish **your aesthetic preference** from **a requirement or an observed user need**. - **What would make me reconsider?** Identify something you check through **a prototype, feedback, or actual use**. - **REMEMBER:** **You can keep your visual sensibility while changing how you apply it.** A **dense screen still needs careful typography, grouping and emphasis.** The craft lies in **making the necessary information manageable.** - ☐ Take your own favourite default (spacious/minimal/subtle) and write down the case where it is the wrong answer. ### 2.8 Start with the screen you're working on - **REMEMBER:** **You don't need to collect another hundred references before practising.** - **PRACTICE — the minimal loop:** - **Name what bothers you.** *"The screen feels crowded"* becomes *"the filters and results compete for attention."* - **Make a change for a reason.** Decide **what should become easier to notice, understand, or do**. - **Check the result.** Use **realistic content**, seek **focused feedback**, and **test with users where the question requires it**. - **Keep a short note about what you learned — especially when your first instinct was wrong.** - **REMEMBER:** That gives you **something to bring into the next project**: a clearer understanding of **which choices work, under what conditions, and why you'd make them again.** --- ## 3. The Fastest Way to Develop Visual Taste as a Designer **Omoraiyewa Samuel · Bootcamp · Jun 2026 · ~974 words** [medium.com/design-bootcamp/the-fastest-way-to-develop-visual-taste-as-a-designer-85ef6d905034](https://medium.com/design-bootcamp/the-fastest-way-to-develop-visual-taste-as-a-designer-85ef6d905034) **Context he sets:** **AI generates interfaces on demand**, and **beginner designers use those outputs as references**. What makes you stand out is **your taste**. Every designer he has watched develop quickly had one thing in common: **they were all inspired by great work** — and that has always been how taste gets built. ### 3.1 What taste actually is - **REMEMBER:** **Taste is your judgment** — the choices you make for the work in front of you: **what font, layout and pattern; what color, spacing; what to emphasise and what to strip away.** - **REMEMBER:** **Every decision you make while designing is an expression of taste**, and **the quality of those decisions is what separates work that feels considered from work that feels assembled.** ### 3.2 The three layers of taste - **REMEMBER — Layer 1, surface:** the **immediate reaction** when you look at something and **know it works, or know something is off, before you can explain why**. This is the layer most people mean by "has taste". **It comes with time and repeated exposure to good work.** - **REMEMBER — Layer 2, context:** **good taste reads the room.** A choice that works beautifully for **one brand, one audience, one cultural moment** can **fail completely in another**. Contextual taste is what makes a design **feel right for the situation, not just visually clean in isolation**. - **REMEMBER — Layer 3, visual:** **the execution layer** — knowing **which typeface pairing carries the right tone**, **how much negative space is actually enough**, and **which color communicates the intended emotion without announcing it**. **This is where taste becomes visible.** - **IMPLEMENT:** Diagnose your own weak layer: surface (can't articulate), context (clean but wrong for the audience), or visual (right idea, poor execution). Train the weak one deliberately. ### 3.3 Why borrowed taste stops working - **REMEMBER:** You can **scroll through Pinterest, find something that looks impressive, and recreate the aesthetic.** That **produces work but does not taste good over time.** - **REMEMBER:** **Borrowed taste is a starting point, not a destination.** The problem with building your eye **entirely on curated feeds** is that **you develop a sensitivity to what looks good on those feeds, not a judgment for what works in context.** - **REMEMBER:** **The moment a design problem steps outside familiar territory, you run out of answers.** - **REMEMBER — what real taste is made of (four ingredients):** **curation across a wide range**, **knowledge of why things work**, **practice making decisions**, and **the experience of seeing those decisions play out**. **None of those are going to come from collecting and reposting inspiration.** ### 3.4 Practice one — study references with a question, not just your eyes - **REMEMBER:** Most people **look at references the way they scroll social media**: something registers as good, they save it, and they move on. **Nothing transfers.** - **PRACTICE — pick one question and look THROUGH it:** - **Why does the hierarchy work here?** - **What makes this feel trustworthy?** - **Where is the designer directing attention and how?** - **What would this look like if the primary action were less obvious?** - **REMEMBER — three sources worth building the habit around:** - **Mobbin** — real production mobile and web UI. - **Layers.to** — curated design work **with context**. - **Your mobile app store itself** — **open apps you use every day and look at them as a designer.** The products that **feel effortless** almost always have **decades of considered micro-decisions** behind them, **visible if you know what to look for.** - ☐ Pick the one question you never ask, and use only that question for a week of reference study. ### 3.5 Practice two — reverse-engineer what you admire - **REMEMBER:** Looking is one layer; **rebuilding is the next logical step.** - **PRACTICE:** **Pick a screen from a product you think is exceptionally well designed. Recreate it in Figma to understand it.** - **REMEMBER — pay attention to every decision that comes up as you build:** the **exact spacing values**, the **type-scale relationship**, **where the designer chose restraint and where they did not**, and **how the component structure creates consistency across states you cannot even see in the screenshot**. - **REMEMBER:** **You will make decisions during rebuilding that the original designer also had to make.** When **yours do not match**, you are **learning**. When **yours actually works better in a specific way**, you are **developing judgment**. - **REMEMBER — flows are especially valuable:** find a well-designed **onboarding sequence, checkout flow, or settings screen** in an app with a **strong design reputation**. **Map the full flow: every state, every transition, every piece of copy.** - **PRACTICE — two questions per step:** **What did the designer assume about the user at each step?** and **What would happen if that assumption were wrong?** - **REMEMBER:** That kind of analysis **builds an instinct for interaction design that no tutorial covers**, because **it comes from studying real decisions made under real constraints.** ### 3.6 Why companies actually hire for taste - **REMEMBER:** **Over the last year, most of the work that came his way arrived because of taste.** **No whiteboarding sessions and lengthy project explanations** — **someone wanted his thinking and his judgment for their product.** - **REMEMBER:** The ability to **make the right call**, **know what fits the brief**, **know what to remove**, and **know what will land with the audience** is **what makes a designer valuable at a level that tools cannot replicate.** ### 3.7 The honest timeline - **REMEMBER:** **Taste takes time. No shortcut changes that.** - **REMEMBER:** **The work that grows your taste is the work you find slightly uncomfortable.** The moment you **have to make a decision you are not sure about, without anything to generate an answer for you**, is **the moment your judgment is actually developing.** - **REMEMBER:** **Do that enough times, and it starts to feel like instinct. That is when you know it is working.** - **IMPLEMENT:** Keep a deliberate "no-generate" slot each week: one decision made entirely from your own judgment, then checked against reality. --- ## 4. What "Good Taste" in UI Design Actually Means **Heshan.Design · Feb 2026 · ~631 words** [medium.com/@heshan.design/what-good-taste-in-ui-design-actually-means-2b52ffc4b3c4](https://medium.com/@heshan.design/what-good-taste-in-ui-design-actually-means-2b52ffc4b3c4) **The article to reread whenever you catch yourself adding instead of deciding.** It is the shortest and, per-word, the sharpest of the five. ### 4.1 What it's *not* - **REMEMBER — the opening question:** *"That designer has good taste."* But what does that actually mean? **Using the right colors? Following trends? Copying great designs? Having a natural talent?** **Not really.** - **REMEMBER:** In UI design, **good taste is not a feeling. It's a skill — and it can be trained.** - **REMEMBER — good taste does not mean:** **fancy visuals, trendy gradients, complex animations, Dribbble-level polish.** - **REMEMBER — a stylish UI can still be:** **confusing, inconsistent, hard to use, overdesigned.** - **REMEMBER:** **Good taste is not about how much you add. It's about knowing what not to add.** ### 4.2 Taste is judgment - **REMEMBER:** At its core, **good taste is decision-making** — the ability to answer: - **Does this element earn its place?** - **Is this helping the user or just decorating?** - **Is this clear enough without explanation?** - **Is this consistent with the rest of the product?** - **REMEMBER:** **Designers with good taste don't guess. They judge calmly and clearly.** - **PRACTICE:** Run the four questions on every element you're tempted to add, before adding it. ### 4.3 Restraint - **REMEMBER:** **Most professional UI looks simple because someone decided to stop.** - **REMEMBER — restraint, concretely:** **not using every color**, **not using every font weight**, **not adding icons everywhere**, **not animating everything**. - **REMEMBER:** **Restraint is a design skill.** **If everything is loud, nothing feels important.** ### 4.4 Hierarchy - **REMEMBER:** Designers with good taste are **obsessed with one thing: what should users notice first?** - **REMEMBER — what they care deeply about:** **visual hierarchy, spacing, alignment, typography scale, contrast.** - **REMEMBER — the test:** **even without colors or images, their layouts still make sense.** **That's not talent. That's trained taste.** ### 4.5 Why beginners struggle - **REMEMBER:** Most beginners **think taste comes first**. In reality, **taste comes after exposure and mistakes.** - **REMEMBER — beginner behaviours:** **add too much, over-style, fear empty space, want to impress.** - **REMEMBER — experienced behaviours:** **remove things, simplify, trust white space, design for clarity.** - **REMEMBER:** **Good taste grows when you stop trying to impress other designers — and start serving users.** ### 4.6 How good taste develops over time - **REMEMBER — the sources:** **seeing many bad designs, fixing confusing screens, watching users struggle, making wrong decisions, learning why something didn't work.** - **REMEMBER:** **It's pattern recognition.** After a while **you just know**: *"This feels unnecessary." "This will confuse users." "This is doing too much."* **That's taste.** ### 4.7 Copying doesn't build taste - **REMEMBER:** **Copying Dribbble shots can help your visuals — but not your taste.** - **REMEMBER — taste improves when you:** **redesign bad products, explain your decisions, compare before vs after, ask "why" instead of "how".** - **REMEMBER:** **If you can't explain why a design is good, you're borrowing taste — not building it.** - **IMPLEMENT:** Redesign one bad screen a month (a login, a settings page, a checkout step) and publish the before/after with the reasoning. ### 4.8 The practical exercise - **PRACTICE — the 5-move drill:** 1. **Open a random app.** 2. **Look at one screen.** 3. Ask: **What's unnecessary here? What's unclear? What's competing for attention?** 4. **Remove or simplify one thing.** 5. **Explain why.** - **REMEMBER:** **Do this often. Taste grows through intentional observation, not inspiration feeds.** ### 4.9 Senior designers aren't more creative - **REMEMBER:** **They're more selective.** **They don't design more. They design less, better.** - **REMEMBER:** That is why their work feels **calm, confident, professional, "obvious" in hindsight**. - **REMEMBER:** **That "obvious" feeling? That's good taste.** ### 4.10 Final thought (and how to know you're improving) - **REMEMBER:** Good taste in UI design **is not a gift**. It's **clear thinking, strong judgment, visual restraint, user-first decisions**. - **REMEMBER — the self-check:** **If your UI feels quieter, clearer, and easier — your taste is improving.** --- ## 5. 10 Tips on How to Develop a Sense of Taste in Design and Other **Kseniia Vertunova · Mar 2023 · ~738 words** [medium.com/@izh.open.yoga/10-tips-on-how-to-develop-a-sense-of-taste-in-design-and-other-41b2dfb24707](https://medium.com/@izh.open.yoga/10-tips-on-how-to-develop-a-sense-of-taste-in-design-and-other-41b2dfb24707) **The graphic-design angle** — 10 years in, and she gives the concept a name that the other four articles describe but never label. ### 5.1 Nasmotrennost' — the concept worth stealing - **REMEMBER:** She has been **a graphic designer for 10 years**, **developing her taste for a long time**. **The most important thing in this field is to have a look.** - **REMEMBER:** In Russian it is called **"насмотренность" (nasmotrennost')** — from **"на" (on)** and **"смотреть" (to look at)**, so the whole word is a noun meaning roughly **"having-looked-at-ness"**. - **REMEMBER — definition in her words:** **to constantly consume/view quality and aesthetic content which is highly appreciated.** - **IMPLEMENT:** Adopt the word. It distinguishes *volume of quality looking* from *collecting inspiration* — the distinction the other articles make with "borrowed taste" and "convergence trap". ### 5.2 The ten tips, in full - **REMEMBER — 1. Study design and art history.** Understanding the history of design and art can help you develop your sense of taste. **By studying different styles and movements, you can learn to appreciate different forms of design and art, and identify what appeals to you.** - **REMEMBER — 2. Follow design and art blogs and social media accounts.** This **exposes you to a wide range of styles** and helps you **stay up to date with the latest trends.** - **REMEMBER — 3. Attend design and art exhibitions.** **See different styles and techniques in person.** **Take note of what you like and don't like**, and **try to identify what makes certain pieces stand out.** - **REMEMBER — 4. Take a design or art class.** Learn **the basics of design and art principles**, and **get a foundation to build on** as you develop your own sense of taste. - **REMEMBER — 5. Read design and art books and magazines.** Learn more about **different styles and techniques**, and **get inspiration for your own work.** - **REMEMBER — 6. Practice design and art.** **As you create your own designs and artwork, you'll learn what works and what doesn't, and develop your own unique style.** - **REMEMBER — 7. Get feedback from others.** Feedback helps you **see your work from a different perspective** and **identify areas for improvement**. **Be open to constructive criticism and use it to refine your work.** - **REMEMBER — 8. Experiment with different styles.** Try **working in different mediums, using different colors and textures, and experimenting with different design styles** — it helps you **find your own unique style**. - **REMEMBER — 9. Trust your instincts.** Ultimately it's about **trusting your instincts and following what feels right to you**. **Don't be afraid to take risks and try new things.** - **REMEMBER — 10. Keep learning.** Developing taste is **an ongoing process**. **Keep learning, experimenting and practicing** and you'll **continue to refine your skills and develop your own unique style.** ### 5.3 The shortcut she actually recommends - **REMEMBER:** Those ten points are **great ideals**, but **it is actually much simpler than that.** - **PRACTICE:** **Set yourself up on Pinterest, and before implementing an idea, start doing a project-order [mini] research — look up relevant work on the topic.** - **REMEMBER — the other channels named:** **Behance, Dribbble, or any other publishers and channels that publish highly appreciated works**, and see **works similar to what you're looking for**. - **REMEMBER:** **A sense of taste is a skill that can be trained.** **If you're constantly looking at what's beautiful, you start to understand what's beautiful and what's tasteless. Your own confidence in it grows. The more you look, the more you know.** - ☐ Before your next project or feature, do a 30-minute "mini research" pass on the exact thing you're designing, and note the 3 references you'll actually borrow a decision from. --- # Part 2 — What all five agree on (the consolidated meaning) Read together, the five articles collapse into one model. This is the part to memorise. ### 2.1 The single definition > **Taste = trained, explainable judgment about what to include, what to remove, and why — developed by volume of looking, volume of making, and the discipline of articulating the difference.** All five arrive here independently: Gupta ("trained judgment… explain why in plain words"), Samuel ("taste is your judgment"), Heshan ("good taste is decision-making… not a feeling, a skill"), Life Considered ("notice what works, understand why, learn when a different approach serves the user better"), Vertunova ("a skill that can be trained… the more you look, the more you know"). ### 2.2 The four ingredients (Samuel's list, corroborated everywhere) 1. **Curation across a wide range** — not one feed, not only apps (Gupta: look outside software; Life Considered: magazines, menus, signage; Vertunova: exhibitions, art history). 2. **Knowledge of why things work** — the three-line note, the one-question study, the feedback four. 3. **Practice making decisions** — handrebuilds, live-project loops, redesigns of bad screens. 4. **The experience of seeing decisions play out** — user testing, before/after comparisons, the judgment log, reopening old work. **None of these come from collecting or reposting inspiration.** That's the sentence all five endorse in some form. ### 2.3 The five repeatable mechanisms | Mechanism | What it does | Named by | |---|---|---| | **Volume of quality looking** ("nasmotrennost'") | Builds the surface layer — instant "this works / something's off" | Vertunova (core), Gupta (habit 1), Samuel (layer 1), Heshan (exposure) | | **Articulation** (write the why) | Converts feeling into transferable, repeatable judgment; closes the vocabulary gap | Gupta (habit 2), Samuel (question-first study), Life Considered (capture template), Heshan (explain or you're borrowing) | | **Reconstruction** (rebuild by hand) | Exposes the decisions you didn't see while looking | Gupta (habit 3), Samuel (practice 2) | | **Critique cadence** (daily, incl. AI) | Turns passive taste into edits and rules; builds pattern recognition | Gupta (habit 4), Heshan (5-move drill), Samuel (layer diagnosis) | | **Out-of-domain input** | Escapes the convergence trap; supplies non-average ideas | Gupta (habit 5), Life Considered (study more than interfaces), Vertunova (exhibitions, art history) | ### 2.4 The shared psychology - **The gap is normal:** eye ahead of hands (Gupta via Ira Glass; Samuel's "honest timeline"; Heshan's "beginners vs experienced"). - **Discomfort is the signal you're training:** Samuel — *the work that grows your taste is the work you find slightly uncomfortable*. Heshan — taste comes **after exposure and mistakes**. Life Considered — keep a note **especially when your first instinct was wrong**. - **Restraint is the visible output of judgment:** Heshan's whole article; Gupta's *crowded dashboard can have great taste if every row earns its place*; Life Considered's *start with grouping and spacing before colours, icons and typography*. - **Serving users beats impressing designers:** Heshan states it directly; Life Considered's "whose preference is driving this?"; Gupta's "taste is not a shortcut around research". ### 2.5 What "improvement" actually looks like (a testable definition) All five accept the same evidence of progress — **your language changes**: - Gupta: **reread week-1 notes and the difference in how you describe things is the proof your eye moved.** - Life Considered: **"every card has equal emphasis, so it's hard to tell what needs attention"** is progress over **"I don't like it anymore"**. - Heshan: **your UI feels quieter, clearer, easier.** - Samuel: **your decisions during a rebuild start matching — or beating — the original designer's.** - Vertunova: **your confidence in the beautiful/tasteless call grows.** --- # Part 3 — Where they disagree (and how to hold the tension) | Tension | Position A | Position B | How to hold both | |---|---|---|---| | **Instinct vs articulation** | Vertunova: *trust your instincts, follow what feels right.* | Gupta/Heshan: *if you can't explain it, you're borrowing taste.* | Trust instinct as a **signal**, then force it through words before acting. An unexplained instinct is a hypothesis, not a decision. | | **Copying** | Gupta/Samuel: **handrebuild is a core drill.** | Heshan: **copying doesn't build taste.** | Copy **to understand decisions**, never to publish the artefact. Copying + explanation = taste; copying alone = visual competence. | | **Breadth vs depth of references** | Vertunova: **consume constantly, broadly.** | Gupta: **save fewer things**; Samuel: borrowed taste stalls. | **Look broadly, save narrowly.** High volume of looking, low volume of saving, and every save carries a question. | | **Minimalism** | Common reflex: simple = tasteful. | Heshan: **knowing what not to add**; Gupta: **a crowded dashboard can have great taste.** | Restraint is a **decision quality**, not a density setting. Dense is fine when every row earns its place (Life Considered's stock-management screen is the failure case: clean, but staff open every row). | | **Speed** | Samuel: *fastest way* to develop visual taste. | All five: **no shortcut on time.** | Fast **method** (start today, on the file that's open) — slow **timeline** (weeks to months). | --- # Part 4 — Coverage matrix (every practice × who teaches it) | Practice | Gupta | Life Considered | Samuel | Heshan | Vertunova | |---|---|---|---|---|---| | Question-named reference collections | ● | ○ | ● | – | ○ | | Three-line "why" note | ● | ● | ● | ○ | – | | One-question-per-reference study | ○ | ● | ● | – | – | | Handrebuild / reverse-engineer in Figma | ● | – | ● | – | – | | Map a full flow (states, transitions, copy) | – | – | ● | – | – | | Daily 10-minute critique | ● | ○ | – | ● | – | | Critique AI output specifically | ● | – | ○ | – | – | | Running rules note → personal style guide | ● | – | – | ○ | – | | Out-of-domain references (magazines, signage, art) | ● | ● | – | – | ● | | Redesign bad products + compare before/after | ○ | ● | – | ● | – | | Feedback with the four questions | – | ● | – | – | ○ | | Tradeoff accounting ("what did I sacrifice?") | – | ● | – | – | – | | Live-project practice loop | – | ● | – | ○ | ● | | Judgment log / expectations vs outcomes | – | ● | – | – | – | | Restraint + hierarchy audit | ○ | ○ | ○ | ● | – | | Deliberate uncomfortable decisions (no AI) | ○ | – | ● | – | – | | Structured 4-week programme | ● | – | – | – | – | | 10 tips / long-term curriculum | – | – | – | – | ● | ● = core to that article · ○ = mentioned/secondary · – = not covered **Read the matrix as a training split, not a scoreboard.** Every row is a distinct rep; if six rows are unchecked in your week, you're practising looking, not taste. --- # Part 5 — The reusable toolbox (verbatim templates) ### 5.1 The reference note (Gupta) + the out-of-domain capture (Life Considered), merged ``` URL / photo: What I see (spacing, type, order, tone): ............ Why it works (the decision, not the vibe): .......... What I would steal / apply (name ONE decision): ..... [out-of-domain only] What did I notice first (size, colour, position, wording)? ... What did the design help me do (find, choose, begin)? ....... Where did I hesitate (missing, unclear, competing)? ......... Source article: ``` ### 5.2 The feedback request (Life Considered) ``` Context: Who uses this screen, and for what task. Constraint: What they need to do / what can't change. Unsure: The one decision I want your judgment on. Ask them: 1. What do you notice first? Does it match what matters for the task? 2. What is difficult to scan or understand? (point to a specific area) 3. What would you change first? What would that improve? 4. What might that change make worse? ``` ### 5.3 The judgment log (Life Considered) ``` Date: Problem named in one sentence ("filters and results compete for attention"): What I changed (grouping/spacing first): What I sacrificed (the tradeoff accepted): What I expected / what happened / what I'd do differently: Still needs checking via (prototype | feedback | user test): ``` ### 5.4 The rules note → personal style guide (Gupta) ``` # My rules - Rule: ... (added when I caught myself repeating this observation) - Test where it broke: ... - Counter-example (when it's wrong): ... ← from Life Considered's "whose preference is driving this?" ``` ### 5.5 The one-question study card (Samuel) ``` Reference: The ONE question I'm looking through: Why does the hierarchy work here? What makes this feel trustworthy? Where is the designer directing attention, and how? What would this look like if the primary action were less obvious? What I found (decision-level): ``` ### 5.6 The four judgment gates before adding anything (Heshan) ``` ☐ Does this element earn its place? ☐ Is this helping the user or just decorating? ☐ Is this clear enough without explanation? ☐ Is this consistent with the rest of the product? ``` ### 5.7 The four context questions before committing a direction (Life Considered) ``` ☐ What task am I making easier? (benefit beyond "it looks cleaner") ☐ What am I making less visible? (do people need that to decide or act?) ☐ Whose preference is driving this? (mine vs requirement vs observed need) ☐ What would make me reconsider? (prototype / feedback / actual use) ``` --- # Part 6 — The practice system (four-week sprint + ongoing cadence) ### 6.1 The 4-week sprint — Gupta's structure, expanded into tickable days **Daily budget: 20–30 minutes.** Same slot every day; the plan fails on irregularity, not on effort. **Week 1 — collect and name.** *Output: ~35 notes + a shared vocabulary.* - ☐ Day 1–5: save **five screens a day**; write the **three lines** under each. - ☐ Day 3: rename **one existing collection** into a question ("How do good apps handle empty states?"). - ☐ Day 5: reread all 25 notes — mark the three where you wrote a **vibe** instead of a **decision**, and rewrite them. - ☐ Day 7: pick the **one study question** you never ask (Samuel) and pin it to your notes tool. **Week 2 — copy.** *Output: 3 handrebuilds + 6 rules.* - ☐ Rebuild **three screens by hand**, one every 2–3 days. **No plugins, no tracing.** - ☐ Before each: **write three guesses about why it looks good.** - ☐ After each: **check which guesses were wrong** (that's the lesson) and **write down two rules you found.** - ☐ Change **one thing on purpose** (type scale or density) and answer better/worse. - ☐ For one of the three, choose a **flow** instead of a screen: map **every state, transition and piece of copy**; at each step ask **what did the designer assume, and what breaks if that's wrong?** **Week 3 — critique.** *Output: 7 critiques + a rules note in your own words.* - ☐ **Ten minutes a day**, one screen: *what works / what does not / what would I change first.* Say it aloud or write it. - ☐ **Half** on human-made work, **half** on AI-generated screens: *is every section the same weight? is the copy vague? does the hierarchy tell you where to look first?* - ☐ Add **a rule to the running note every time you catch yourself saying the same thing twice.** - ☐ Use Heshan's variant on two of them: *what's unnecessary / unclear / competing — remove or simplify one thing and explain why.* **Week 4 — look outside and make.** *Output: 10 outside references + 1 designed-by-rules screen + a comparison.* - ☐ Collect **ten references from outside software** (posters, packaging, street signs, maps, film title cards, buildings) and run the **four capture questions** on each. - ☐ **Design one screen using only your written rules.** - ☐ Generate **the same brief with an AI tool** and **compare**: where did it converge on the average, and where did your rules beat it? - ☐ **Reread your week-1 notes.** The change in how you describe things **is the proof your eye moved.** ### 6.2 The ongoing cadence (after the sprint) | Cadence | Rep | Source | |---|---|---| | **Daily, 10 min** | One critique: what works / what doesn't / what I'd change first. Half the week on AI output. | Gupta, Heshan | | **Daily, 2 min** | Add a rule to the running note when you repeat an observation. | Gupta | | **Per reference saved** | Three lines + one named decision. No bare saves. | Gupta, Life Considered, Samuel | | **Per project** | One "mini research" pass on the exact thing being designed (Pinterest/Behance/Dribbble/Mobbin) before implementing. | Vertunova | | **Per week** | One decision made **without** AI generation, then checked against reality. | Samuel | | **Per 2 weeks** | One handrebuild, three guesses written first. | Gupta, Samuel | | **Monthly** | Redesign one **bad** screen (a login, settings page, checkout step) and publish before/after **with the reasoning**. | Heshan | | **Monthly** | Run the Life Considered loop on the part of your live project that bothers you most; keep the judgment log. | Life Considered | | **Quarterly** | Reopen a design from months ago; **write what you notice before fixing it**; check the four maturity signals. | Life Considered | | **Quarterly** | Coverage-matrix audit: which rep haven't I done in 12 weeks? | This document | ### 6.3 The 10-minute critique, three versions (use all three in rotation) 1. **Gupta (daily default):** what works · what does not · what would I change first. 2. **Heshan (when you're about to add something):** what's unnecessary · what's unclear · what's competing for attention → **remove or simplify one thing, explain why.** 3. **Samuel (when studying a reference):** pick one question — hierarchy? trust? attention direction? what if the primary action were less obvious? — and look *only* through it. --- # Part 7 — Master checklist ### 7.1 REMEMBER (the principles, condensed to load-bearing statements) - ☐ Taste is **trained judgment you can explain**, not style, trend, gear, or talent. - ☐ **If you can't explain it, you can't repeat it, and you can't teach a tool to do it.** - ☐ Your **eye starts ahead of your hands**; the gap closes with volume of work, not waiting. - ☐ Taste is **not minimalism** — a crowded dashboard can have great taste if every row earns its place. - ☐ In B2B, taste shows in **tables, forms, empty states**, not hero images. - ☐ AI makes decent screens **cheap**; the scarce skill is **choosing and knowing what to change**. - ☐ The tool returns **the average** — the **convergence trap**; only your eye pushes away from it. - ☐ The **vocabulary gap** is a measurable failure mode: vague words in, average out. - ☐ NN/g's four automation-resistant skills: **curated taste, research-informed context, critical thinking, careful judgment** — taste is one of four, not a substitute. - ☐ **Good taste with no user understanding = a pretty wrong answer.** - ☐ Taste has **three layers** — surface, context, visual. Train the weak one. - ☐ **Borrowed taste stalls** the moment a brief leaves familiar territory. - ☐ **Restraint is a skill**: most professional UI looks simple **because someone decided to stop**. - ☐ **If everything is loud, nothing feels important.** - ☐ The one question of hierarchy: **what should users notice first?** - ☐ **Senior designers aren't more creative — they're more selective. They design less, better.** - ☐ **"Obvious in hindsight" is what taste feels like from the outside.** - ☐ Good taste grows when you **stop impressing designers and start serving users**. - ☐ **Discomfort is the training signal** — the decision you can't generate an answer for is the rep. - ☐ **Nasmotrennost'**: constantly consume quality work; the more you look, the more you know. - ☐ Improvement shows up as **changed language**, not as a changed feed. ### 7.2 PRACTICE (the reps) - ☐ Three-line note under every save. - ☐ Question-named collections. - ☐ One-question reference study (rotate the four questions weekly). - ☐ Handrebuild a screen, three guesses first, two rules after. - ☐ Map a full flow: every state, transition, copy; what did the designer assume? - ☐ Daily 10-minute critique — half on AI output. - ☐ Heshan's five-move drill: random app → one screen → unnecessary/unclear/competing → remove one thing → explain why. - ☐ Redesign a bad screen; compare before/after; explain decisions. - ☐ Out-of-domain reference capture, four questions each. - ☐ Live-project loop: name the problem → grouping/spacing first → realistic content → explain → check the sacrifice. - ☐ Feedback request with the four questions, including *what might that change make worse?* - ☐ Judgment log: expected / happened / would do differently. - ☐ Reopen old work and **write before fixing**. - ☐ One decision a week made with **no AI generation**. - ☐ Month-end: reread your notes and mark where your language sharpened. ### 7.3 IMPLEMENT (changes to how you work) - ☐ Make the three-line note **mandatory** in your reference tool (Mobbin collections, or a notes field). - ☐ Add a **"Why this works"** field to your design review template, with the four judgment gates. - ☐ Replace **"looks good"** in reviews with the four feedback questions — and ask for **what would get worse**. - ☐ Add **"What did we sacrifice?"** to every design decision record (Life Considered's tradeoff check). - ☐ Add **"Whose preference is this?"** to any reduction that hides information — the stock-management test. - ☐ Keep the **running rules note** as a living style guide in your own words; review it monthly. - ☐ Institute a **monthly bad-screen redesign** ritual (45 min) and keep the before/after with reasoning. - ☐ Before each project: **30-minute mini research** on the exact artefact type (Vertunova). - ☐ Instrument your **vocabulary gap**: log the vague prompts you write to AI tools, then translate each into decision language. - ☐ Run the **4-week sprint** once per quarter; `git commit` the notes so quarter-over-quarter language is comparable. - ☐ Run the **coverage-matrix audit** quarterly: which of the 18 practices did you skip? --- # Part 8 — Failure modes, named | Failure mode | What it looks like | Who names it | The counter | |---|---|---|---| | **Convergence trap** | AI output drifts to generic and polished; you accept it because it looks fine | Gupta (Berkeley study) | Critique AI output daily; design one screen only from your written rules | | **Vocabulary gap** | You know it's wrong but write "make it cleaner" | Gupta (Berkeley study) | Three-line notes; decision-level language; log vague prompts | | **Borrowed taste** | Sensitivity to what looks good in feeds, no judgment for context | Samuel; Heshan ("if you can't explain why, you're borrowing taste") | Question-first study; rebuild by hand; explain decisions | | **Inspiration-feed scrolling** | Save, move on, nothing transfers | Samuel | One question per reference; four capture questions | | **Style ≠ taste** | Fancy visuals, gradients, complex animation, Dribbble polish on a confusing UI | Heshan | Four judgment gates; hierarchy test without colour/images | | **Impress-the-peer reflex** | Adding to be admired by designers, not served by users | Heshan | Serve users; ask "helping or decorating?" | | **Aesthetic preference overriding need** | Hiding secondary info "to be cleaner" until staff open every row | Life Considered | The four context questions; "what am I making less visible?" | | **"Looks good" feedback** | Reassurance instead of information | Life Considered | The four feedback questions; ask what would get worse | | **Taste as research substitute** | Beautiful and wrong | Gupta (NN/g) | Taste for feel/look; users for whether it works | | **Waiting for the gap to close** | Paralysis because eye > hands | Gupta (Ira Glass) | Volume of work; ship the uncomfortable decision | --- # Part 9 — Limits and honest caveats (kept, not smoothed over) - **Taste is not a shortcut around research.** Use taste to judge how something feels and looks; use users to judge whether it works. (Gupta, citing NN/g's four skills.) - **No shortcut on time.** Taste cannot be faked in a weekend; the only accelerant is method, not shortcut. (Samuel, Gupta, Vertunova all say some version.) - **Trusting instinct is a late-stage instruction, not an early one.** Vertunova's "trust your instincts" arrives after ten years of volume — pair it with the explanation requirement, or it becomes unexamined preference (see Life Considered's "favourite approach doesn't fit"). - **Your three-layer weakness is specific.** Being strong at surface and weak at context is a different training plan from the reverse; diagnose before drilling. (Samuel.) - **Reading about taste is not practising it.** Every article here ends in an exercise for a reason — this document is a plan, not progress. --- # Sources 1. Nishkarsh Gupta, *How to Develop Design Taste in the AI Era: A 4-Week Plan for UX Designers*, ILLUMINATION, Oct 2026 — [link](https://medium.com/illumination/design-taste-new-skill-how-to-build-it-a46d424f3173) 2. Life Considered, *Everyone Says Designers Need "Taste". How Do You Actually Develop It?*, Muzli — Design Inspiration, Oct 2026 — [link](https://medium.com/muzli-design-inspiration/everyone-says-designers-need-taste-how-do-you-actually-develop-it-c8c09f0e2605) 3. Omoraiyewa Samuel, *The Fastest Way to Develop Visual Taste as a Designer*, Bootcamp, Jun 2026 — [link](https://medium.com/design-bootcamp/the-fastest-way-to-develop-visual-taste-as-a-designer-85ef6d905034) 4. Heshan.Design, *What "Good Taste" in UI Design Actually Means*, Feb 2026 — [link](https://medium.com/@heshan.design/what-good-taste-in-ui-design-actually-means-2b52ffc4b3c4) 5. Kseniia Vertunova, *10 Tips on How to Develop a Sense of Taste in Design and Other*, Mar 2023 — [link](https://medium.com/@izh.open.yoga/10-tips-on-how-to-develop-a-sense-of-taste-in-design-and-other-41b2dfb24707) **Secondary sources cited inside the articles:** NN/g *State of UX 2026* (Kate Moran, Raluca Budiu, Sarah Gibbons, Jan 2026); UC Berkeley School of Information 2026 final project on prompt-based interface tools (13 UX designers, Lovable); Ira Glass on the creative gap. Full unedited article text: `medium-design-taste-full-articles.md` (compiled alongside this document).