Taste: The same bottleneck in both the academic and entrepreneurial circles

Writing: Yajin

Recently, I encountered a few things that gave me some insights into the trending word “taste” lately, so I’m sharing my thoughts with everyone.

1. Full resume, shallow understanding

This week, I interviewed an undergraduate applicant for our team.

His resume looked quite good. He had participated in three research projects and published a paper. For an undergrad, this output already surpasses many master’s students.

At the start of the interview, I asked him what motivated his first project. He gave a somewhat vague answer. I inquired about technical details; he could explain what he did but couldn’t clarify why he did it. What problem did this work solve? How does it differ fundamentally from previous methods? He couldn’t answer.

The second project was similar.

By the third project, I roughly understood. This student did a lot, but never truly understood any of it. His research experience wasn’t “I’m interested in a problem and delve deep,” but more like the common “adding research experience” on Xiaohongshu—participate when there’s a chance, move on after completing it, and just add another line to the resume, turning research into a points game.

2. Another kind of “cheater”

Around the same time, a friend mentioned a phenomenon in the hackathon scene.

There’s a type of participant who attends hackathons everywhere—this weekend at this competition, next weekend at that one. Their resumes are filled with “Won awards at XX hackathon,” but if you look closely, what they do each time is pretty much the same: wrap an AI API, add a UI layer, make a demo. After the competition, the project dies.

My friend calls these people “hackathon cheaters.”

Hearing this term, I suddenly realized it’s the same problem as with the student I interviewed.

On the surface, one is in academia, the other in startups—completely different scenarios, but the core is the same: replacing depth with quantity, understanding with experience, and relying on numbers on the resume instead of real judgment.

This behavior pattern has a more precise name: “shallow experience.”

3. The limit of “shallow experience”

Don’t get me wrong—I’m not saying that accumulating experience is useless. For newcomers, broad attempts can help you understand the overall landscape of a field and find your interests.

But there’s a hard ceiling: it can help you “know what exists,” but not judge “what is worth doing.”

This limit is visible in many places.

There are over 2 million apps on the Apple App Store. According to Business of Apps, nearly a quarter of these apps have fewer than 100 downloads. Developers work hard, but most create something “usable” that no one needs.

The AI tools sector is even more obvious. Over the past two years, a flood of AI wrappers entered the market, doing similar things: wrapping a large model API, adding a UI, solving a vague “improve efficiency” need.

Most are ignored, but a few products survive and thrive. The difference isn’t technical ability or funding; it’s taste.

4. What is Taste?

“Taste” is a difficult word to translate. It encompasses style, aesthetics, judgment—each captures only part of the meaning.

My understanding of taste is: when faced with 100 possible actions, the ability to pick the one truly worth doing.

Steve Jobs once said in a famous 1995 interview: “The only problem with Microsoft is they just have no taste. They have absolutely no taste. And I don’t mean that in a small way. I mean that in a big way, in the sense that they don’t think of original ideas and they don’t bring much culture into their product.”

The key point isn’t about the interface’s appearance. It’s about Microsoft not thinking about what is truly original or culturally meaningful in their products. They can do everything but don’t know what’s most worth doing.

Of course, saying Microsoft is unsuccessful isn’t accurate. Microsoft is very successful commercially, but its product lines feel fragmented. At CUHK, we use Microsoft enterprise solutions, including Microsoft 365, which we also used early on. Honestly, it’s hard to use, hard to describe. Selling B2B products involves many factors beyond the product itself; taste isn’t the only variable here.

Richard Hamming, in his 1986 classic speech “You and Your Research,” told a story. When at Bell Labs, he often asked colleagues three questions at lunch: What is the most important problem in this field? What important problems are you working on? If what you’re doing isn’t important, why do it?

Most people stopped eating with him after the third question.

But Hamming’s logic was clear: doing the right thing is more important than doing things right. “If you do not work on an important problem, it’s unlikely you’ll do important work.”

Some might say, I’m not aiming to be a Hamming-level scholar—what does this have to do with me? Actually, Hamming’s point isn’t only for top scientists. Whether doing research, building products, or choosing a job, the core question is the same: what are you spending your time on?

That’s taste. In academia, taste is the ability to choose the right research problems. In industry, it’s the ability to pick the right product direction.

5. Academic Taste: A Personal Experience

In 2012, we published a paper on Android security at IEEE S&P (a top conference in security).

Looking back, Android security is now a mature field studied for over a decade, but in 2012, it was very different. Android had just been out for a few years, and the academic focus on mobile security was minimal; most researchers still concentrated on traditional PC security.

As a PhD student then, I lacked judgment about the direction. Choosing Android security was my advisor’s taste. He saw smartphones becoming the main computing platform, and security issues would explode with it. This judgment wasn’t obvious at the time—many thought there wasn’t much security worth researching on phones.

It turned out to be the right choice. The paper was cited many times, and more importantly, it established our footing in Android security. Many subsequent works built on this foundation.

Conversely, if my advisor hadn’t had that taste, we might have chased the trending topics, doing what everyone else was doing. Maybe we could have published papers, but probably wouldn’t have had such influence.

This shows the value of academic taste. Choosing the right problem gives your work direction for years. Choosing the wrong problem, no matter how hard you work, just piles up numbers.

6. Industry Taste: Even More Evident in the AI Era

In industry, taste manifests differently: in product selection. Making something “usable” is easy; making something indispensable to users is hard.

The AI era amplifies this issue.

Because AI greatly reduces execution costs. Previously, building a product required a team months; now, a single person can prototype in days using AI. Execution is no longer the bottleneck—judgment is.

This is similar to the App Store situation. Development skills are no longer the barrier; most lack a sense of direction. When everyone can make apps, being able to make apps isn’t a competitive advantage. Knowing what kind of app to make is.

The most direct example is AI productivity tools. In 2024-2025, hundreds of such tools emerged. Most do similar things: call large model APIs, add a UI, solve a vague “improve efficiency” need.

A few early on found the right direction. For example, some teams rethought “how should programming be assisted by AI,” others redefined “what search should look like in the AI era.” Whether these products will succeed is uncertain, but their starting point is taste: choosing what problem to solve, for whom.

7. Where Does Taste Come From?

A natural question: can taste be cultivated or is it innate?

Paul Graham gave a good answer in his essay “Taste for Makers”: taste isn’t just subjective preference; it’s a judgment that can be developed.

He said good design shares traits: simplicity, solving the right problem, appearing effortless but requiring effort behind the scenes. Cultivating taste hinges on “intolerance for ugliness.”

Here’s a seeming contradiction: product people often say “don’t pursue perfection, launch first and iterate,” isn’t that at odds with “intolerance for ugliness”? I think it’s not. Taste involves not settling on the direction; if you pick the wrong problem, no amount of perfect execution helps. But at the execution level, making a rough version quickly to validate the direction is a sign of taste: focusing on judgment rather than polishing something that might not need to be made at all.

Based on my experience and observations, I see several ways to develop taste:

First, expose yourself to a lot of “good” things.

Reading enough excellent papers helps you recognize what’s bad. Using enough good products helps you see what’s good. The starting point of taste is exposure.

Second, work with people who have taste.

My Android security taste came from my advisor. He didn’t explicitly teach me what taste was, but through discussions, I gradually understood how he viewed problems and judged whether a direction was worth pursuing.

Taste is hard to learn from books because it’s a judgment, not knowledge, but it can be transmitted through long-term interactions with people who have it.

That’s why working in good labs and with talented peers is so important. Being around excellent people allows you to grow through communication. Sadly, I’ve seen many people treat talented peers as enemies, blinded by jealousy and losing rational judgment.

Third, deepen your expertise in a field.

The problem with shallow experience is that you’re just a tourist in each domain. Tourists see the sights; locals know which road leads where.

Deep work in a field helps you develop a sense: understanding what the real hard problems are, what surface issues are, what methods are correct, and what are dead ends. That sense is taste.

Fourth, learn to say “no.”

Ultimately, taste is about choosing what not to do. For researchers, it means rejecting topics that can be published but aren’t important. For entrepreneurs, it’s rejecting directions that have market potential but aren’t worth pursuing.

8. Returning to that interview

Back to the interview at the beginning.

The student wasn’t lacking in ability or effort. His problem was that, over the past few years, no one told him (or he didn’t realize) that doing three shallow projects