设计是清晰思考的可见化：Bob Baxley 谈文化、信条，以及软件作为媒介

Baxley recalls a story, which he is not certain of the source of, that Steve Jobs listed Apple itself as the product he was proudest of. He reads this as a bet on the longevity of culture over any product. In his telling, Apple's culture is a way of making decisions, down to a patented pizza box in the cafeteria, driven by the constant question of how the thing you are working on can be a little better.

Asked why so few Apple alumni have built great things elsewhere, Baxley points to his own year and a half at Pinterest. He left Apple on a Friday and started on a Monday, with no time to recalibrate. He says powerful cultures indoctrinate you deeply, and he notes Hiroki Asai's success at Airbnb came after a multi-year gap, a kind of car wash after leaving Apple.

His takeaway is that a new employer hires you for the values of the place you left, not its behaviors. At Apple those values were attention to detail, product excellence and doing everything for the user. He says he expressed them more successfully in ThoughtSpot's culture than in Pinterest's, where he arrived being very direct and fighting hard while the company's posters urged people to "say the hard thing."

Baxley describes senior careers as going out on a branch that gets flimsier, often while family life is also demanding. He argues that people falling off such jobs is the common case and that we are biased toward survivors. He also suggests a startup's job is to grow its founder, and that executives are often outgrown by the pace of a company, which he says feels like failure but is a normal human experience.

To avoid culture clashes, Baxley interviews with founders and listens for a real story about why they believe in design. He describes ThoughtSpot founder Ajeet Singh, who saw the power of design through IDEO engagements early in his career. Baxley says he has never seen design grafted on after a company's founding, and Lenny adds that people in any function should ask whether the founders value their craft.

For Baxley, design is Edward Tufte's "clear thinking made visible," not a team or a visual expression. He describes it as imagining the future you want to live in and taking steps to make it real. That differs from science, which observes, and from engineering, which moves one step at a time toward a known end.

When Lenny describes design as beautiful, intuitive products, Baxley says that is only the part of the iceberg above the waterline. The deeper work is alignment on why the company exists and how vision, mission, tenets and design strategy ladder into execution. Without that, he says, you have bricks scattered across the backyard instead of a wall.

He says that clarity let Apple's online store run with six designers across more than 30 countries, where other companies might need 60 or more. Small teams also make it more likely that minds fuse into one whole. He jokes that you get The Beatles with four people, not eight or 24, and notes the original Mac and the iPhone each had roughly 20 to 24 people on the patent.

Baxley quotes a friend's line that saying a company is design-led does not mean it is designer-led. He wants designers to inhabit a product mindset too, but he insists on respecting role boundaries. He expects product partners to own the roadmap, in his view because they know the customer and the business better, and asks them to leave the design solution to design.

He pushes back on the idea that generative AI means nobody needs teammates. In his view each role requires a mindset that fills a person's head, and "the second baseman doesn't cover first." He says creative tension is where the magic happens, as long as the team still enjoys working together.

His most contrarian view is that design works best as phase zero of engineering. When design works only with product, he says, the pair can cook up things that are hard to build while engineering is brought in too late. At Apple, he says, design always reported to engineering under Steve Jobs. He is clear that every culture is different and that he mainly wants leaders to weigh three options consciously.

If a company will not restructure, he suggests finding a few "creative technologists" in engineering who can sit with ambiguity in early philosophical conversations. He wants makers to be part of the idea from inception, not to give buy-in later. His test is that a few key engineers and product people should fall in love with the work so they can defend and extend it.

Baxley rejects the premise that design takes more time, arguing that a design exists either way and the question is whether it is good. A shared philosophical understanding lets teams ask what a company like theirs would do. He observes that the largest design teams often sit in companies with the most ambiguous cultures, comparing it to how a teenager takes longer to decide what to wear than someone with a settled personal style.

He distinguishes tenets from principles. Words like simple, clear and fast are platitudes because nobody argues for the opposite, whereas tenets are decision-making tools, like choosing paper or plastic once. He illustrates with a story that Steve Jobs told the Keynote team to make ugly presentations hard to produce, focus on cinematic transitions and favor innovation over PowerPoint compatibility.

At ThoughtSpot, where many engineering teams were in India, Baxley says he moved from control to choreography. The team had three tenets: documentation is a failure state, every interaction starts simple and lets users opt into complexity, and the whole product should feel like it came from a single mind. The second was meant to contrast with Tableau, which he says begins with complexity.

For finding your own tenets, he suggests watching for debates that keep recurring with the team digging into two camps, then having that debate once and deciding. He adds that no unopinionated software has been successful. He says he chose the word tenet to avoid fighting over principles, which he calls "apple pie and motherhood."

Baxley frames great product as a moral obligation. In an airport, he says, you can watch people's faces as they struggle with phones, and almost everyone now has hundreds of consequential software interactions. When he asks audiences whether they have had a frustrating software experience that morning, most hands stay up. Each frustration drains a bit of energy, and the work is anonymous, so makers never see who is affected.

He argues that makers should put that emotional energy back into people's lives and that every demand a product makes on the user is a failure on the team's part. People rarely grasp the scale of what they create in Figma that billions will use. Lenny relates this to the motivation he feels when someone clicks to watch an episode.

His practical advice is to go out of your way to watch people use software in the wild: self-checkout lines, ATMs, kiosks, car dashboards, even friends' phones. He compares relying on metrics alone to reading a radio telescope's raw data instead of looking at the night sky. Filmmakers and comedians watch audiences, but software makers have no obvious equivalent.

He suggests watching people use other companies' products rather than your own, because you bring a bias to your own. For a checkout redesign, his team ran a lab session in which participants went through eBay, Williams-Sonoma and Amazon. They learned, for example, that the shipping quote was nearly as important as price.

Baxley says he left Pinterest in 2016 without a plan and, while listening to a Stanford podcast about Silicon Valley, heard a line from Fred Turner's book *From Counterculture to Cyberculture*. It said personal computing began near Stanford because a small group saw software as a new media form on par with movies, music and books. The idea clicked for him: he had been a photographer, a film student and a musician, and had been hunting for his medium until he found software at 27.

He argues that software is a medium because it always evokes emotion, unlike a hammer or a calculator: you feel confused or empowered, your world feels bigger or smaller. Teams can leave that to chance or choose the emotion deliberately. He pushes back on executives who say a visual choice is a matter of opinion, because blue or red will evoke something, so the question is what you want users to feel.

He tells the Toast team that at dinner tonight a grandmother in Ohio will hand over the payment device. Design decides whether she looks like a superhero or her grandson has to take over. Asked about the emotional frame, he says he does not think about what he wants users to do, since that can be exploitative, and instead asks what is right for them.

On metrics, Baxley says he has never seen a successful product use metrics as its driver. He has seen companies treat them as a consequence and a feedback mechanism for judging decisions. He believes prioritizing the right thing for the user leads to good outcomes for the money metrics too.

Asked how to contain the design process, Baxley compares a creative brief to an AI prompt: the more context and specificity you give, the faster the designer finds the right lane. If design feels slow, he says, the brief is probably unclear because the requester is not clear in their own head. He cites vision statements such as Google's organizing the world's information and Disneyland's happiest place on earth as examples of vision that organizes decisions.

He says he never understood Yahoo's founding vision and believes that ambiguity made the company inefficient. He cautions founders that the company is not the product and warns that Slack and Pinterest became companies without a larger vision of the change they wanted to see. Lenny adds that working with AI teaches the same lesson as working with people: when results disappoint, give more context.

One of Baxley's main jobs as a design leader is making product managers better clients. Designers receive ambiguous input but must hand engineering something highly specific, often on a deadline. He uses a film analogy: if the script is weak, storyboards suffer, and ambiguous storyboards waste money on set. The faster design goes the more ambiguity has been removed upstream.

He cautions PMs against handing over a drawn, fully baked solution, since it only tells him what the team will not do. PMs should describe the problem in words and leave room for original ideas. He suggests giving designers constraints like a basketball court, rather than an airport tarmac or a football field.

Baxley's counterintuitive advice is to hold off on drawing. He borrows from art the idea of a primal mark: the first mark on the canvas, which everything afterward responds to. Once something looks even slightly real, people feel uncomfortable with ambiguity and double down on it, shrinking a broad space of possibilities to one tiny option, which is worse when it comes from an AI trained on existing solutions.

He thinks you can stay conceptual and conversational and reach a second, third and fourth idea within a single meeting. He tells of a PM who made a homepage link blue by going around him, because it was not prominent enough. His point was that blue was only the first idea of a non-designer, and there were a hundred ways to solve the actual problem.

On AI prototyping tools, he calls them a production tool for when an idea is ready. He says an interactive idea should eventually be expressed interactively, but fragile ideas squashed by early critique need space, like the little plant in WALL-E. High-resolution mockups also draw feedback about colors and shapes, which he likens to judging a movie's special effects when the story is bad.

At ThoughtSpot his team used chunky "block brain" diagrams and wireframes, which forced conceptual debate and sometimes took weeks. With a strong design system, an agency could then produce high-resolution comps in a day, which surprised PMs. He says the hard part is deciding what you are trying to do. Against the push to prototype everything, he questions whether a faster spaghetti-throwing machine produces better products.

In the AI Corner segment, Baxley says his current job is figuring out what his job is, so he has used ChatGPT as a life coach. One prompt that worked well asked for an outdated mindset he was holding onto. The reply said he was wedded to control, which would not suit writing and speaking, and it produced the phrase "choreography over control."

He then started a fresh project and asked for five questions a day for five days, so it would stop inferring from past chats. He stresses that it is no substitute for a therapist or real coach, since it does not care about him as a human does. Its value, he says, is reflecting back things already in his head.

He explains this with Guy Claxton's idea of the undermind, which processes information before it becomes language. In his compiled-versus-interpreted code analogy, the undermind runs like compiled code while consciousness is slower interpreted code. He thinks the model statistically reflects patterns already in his undermind and puts them into words his conscious mind can recognize.

Baxley, a fan of the Apollo program, retells the story of John Houbolt. After President Kennedy's Rice University speech, which Baxley calls the perfect TED talk at about 18 minutes, nobody at NASA knew how to reach the moon. Options included building one giant rocket for a direct ascent, an Earth-orbit rendezvous, and lunar orbit rendezvous, in which a small, light craft would carry the crew to the surface.

The lunar orbit idea was first theorized by Yuri Kondratyuk in Ukraine around 1918. Houbolt discovered it years later and could not get traction, so he bypassed the hierarchy with a famous memo arguing that going to the moon is a weight problem. Baxley notes that he risked his career, and that NASA eventually adopted the approach.

Baxley draws three lessons. First, they did not know how to do it, and that uncertainty deserves awe. Second, a great idea can sit and wait for its time. Third, ideas need champions who will put themselves on the line, and he urges people not to avoid posting publicly out of fear of self-promotion, because they are advocating for ideas, not themselves.

Baxley recommends three books: *The Elements of Typographic Style* by Robert Bringhurst, which he says opens up the designer mindset; *Zen and the Art of Motorcycle Maintenance*, on quality and cohesive wholes; and Robert Grudin's *Time and the Art of Living*. For TV he enjoyed *Severance* as commentary on the modern workplace, and he recommends *Lawrence of Arabia* as one of the most complete expressions of what film can do.

For favorite products he names his Leica M6 film camera, arguing that your tools change how you show up: an iPhone invites sharing, film invites careful composition, and a digital SLR invites volume. He also praises the habit app Habitica, a role-playing game mixed with a to-do manager, as an example of mashing up conceptual models the way Star Wars mixes cowboy and space genres.

His three recurring quotes are Tufte's "design is clear thinking made visible," Ansel Adams's line that nothing is worse than a brilliant image of a fuzzy concept, and an African proverb about going far together. He ties them to AI tools: it is now easy to produce high-production-value work that means nothing, like fancy potato chips with no nourishment. He also notes Hollywood planted the dystopian baseline for how people see AI.

Asked what he would change as owner of the Warriors, Baxley says a real team cannot depend on a single player, citing the gap when Steph Curry is off the court. He closes by saying that what he asks of listeners is about ideas rather than himself: that software matters because it is made for people on the other side of the glass. He hopes people will take more responsibility for cleaning up a digital world he says is not safe for our kids.