<div className="image-container"><img src="/static/img/crr.png" style="width: 100%; margin-top: 12px" alt="Agnes Martin">
<span className="caption">Reimagine a simplified version of code review on GitHub</span></div>

Code review is evolving. As AI-generated code becomes prevalent, so does the complexity of reviewing it: <i>validation</i> (checking functionality and catching edge cases) and <i>understanding</i> (evaluating deeper implications and compliances) are getting more complex, contextual, and individualized.

So how do we build reflective<sup>1</sup>, not just reactive, review tools? After spending the past month exploring this space, a few thoughts:

<h3>Foundation still matters</h3>

<div className="image-container">
<picture>
  <source srcset="/static/img/crr.webp" type="image/webp" />
  <img src="/static/img/crr.png" style="width: 100%; margin-top: 12px" alt="Agnes Martin" />
</picture>
<span className="caption">Reimagine a simplified version of code review on GitHub</span></div>

Clarity of structure is more important than ever. Any lack of focus and clear hierarchy will only get amplified with more complex, automated tasks. But this also presents an opportunity: we <i>can</i>, and <i>should</i>, create unified<sup>2</sup>, efficient interfaces make both automated and manual work easier to navigate.

<h3>The fluid, dynamic nature</h3>

<div className="image-container">
<picture>
  <source srcset="/static/img/crr-fluid.webp" type="image/webp" />
  <img src="/static/img/crr-fluid.png" style="width: 100%; margin-top: 12px" alt="Agnes Martin" />
</picture>
<span className="caption">Manging tasks, steering actions, and understanding the system.</span></div>

The boundary between generating and understanding is dissolving. People might need to frequently switch between asking questions, steering actions, and defining or managing tasks. Should tools specialize by task? Or converge into a single, adaptable surface? Likely somewhere in between.

<h3>Balancing social and personal dimensions</h3>
What does collaboration look like when automation drives so much <i>personal productivity</i>? Maybe more than ever. We need tools that help users surface problems in complex systems, trace historical decisions, and—crucially—build a shared mental model of how things work.
<br> <br>

<div className="image-container">
  <video autoplay muted loop style="width: 100%; border-radius: 8px; border: 2px solid #e5e7eb;">
    <source src="/static/videos/pr-new.mp4" type="video/mp4" />
    Your browser does not support the video tag.
  </video>
  <span className="caption">A prototype my coworker Skylar Andersen and I built for automatically summarizing changes and breaking down PRs.</span>
</div>

<div class="footnote"><sup>1</sup> In Ted Chiang's story <i>The Truth of Fact, the Truth of Feeling</i>, speech recognition becomes so prevalent that people started losing their ability to write and clearly formulate thoughts. Today, we face a different yet related challenge: content is being generated and regenerated faster than we can process it. Information is everywhere, but understanding is harder to come by. </div>

<div class="footnote"><sup>2</sup> Many so-called "reinventions" of agent interfaces simply repackage well-established design patterns—lists, comparison tables, kanban boards. These structures are effective, but innovation should be purposeful, not driven by novelty or hype.</div>

Code review is evolving. As AI-generated code becomes prevalent, so does the complexity of reviewing it: validation and understanding are getting more complex, contextual, and individualized.

Rethink Code Reviews

<i>This guide summarizes what I've learned in the past year working on early AI experiments at GitHub.</i>

Just as designing for any technical products, understanding how things work under the hood matters. This is for designers new to the space wanting to learn a bit more about the <i>challenges</i> and <i>constraints</i>. It's not a deep dive, but hopefully a solid starting point.

<h3>The core components</h3>

Think of an AI system as a little team responsibile for output quality: the <i>model</i>, the <i>contexts</i>, and the <i>prompts</i>. Improving one usually helps the others. For instance, good prompts often involve context specification, and effective context-gathering requires thoughtful prompt design.

<div style="width: 336px; max-height: 400px; margin: 2rem auto;">
  <svg viewBox="0 0 500 500" width="100%" height="100%" style="max-height: 400px; display: block; margin-left: -12px">
  <style>
    @keyframes dashFlow {
      to {
        stroke-dashoffset: -12;
      }
    }
    .theme-stroke {
      stroke: var(--theme-color, #666);
    }
    .theme-fill {
      fill: var(--text-color, #666);
    }
    .dashed-line {
      animation: dashFlow 1.5s linear infinite;
      stroke: var(--theme-color, #777);
    }
    .arrow {
      stroke: var(--theme-color, #666);
      fill: none;
    }
    text {
      font-family:  -apple-system, BlinkMacSystemFont, "Segoe UI", Roboto, Helvetica, Arial, sans-serif;
      font-weight: 500;
      color: var(--text-color, #666);
    }
  </style>
  <defs>
    <marker id="arrowhead" markerWidth="10" markerHeight="7" 
    refX="9" refY="3.5" orient="auto">
      <polygon points="0 0, 4 3.5, 0 7" fill="#666"/>
    </marker>
  </defs>

  
  <circle cx="250" cy="80" r="60" fill="none" class="theme-stroke" stroke-width="1"/>
  <text x="250" y="80" text-anchor="middle" dominant-baseline="middle" font-size="22" class="theme-fill">Models</text>

  
  <circle cx="250" cy="250" r="60" fill="none" class="theme-stroke" stroke-width="1"/>
  <text x="250" y="250" text-anchor="middle" dominant-baseline="middle" font-size="22" class="theme-fill">Quality</text>

  
  <circle cx="100" cy="420" r="60" fill="none" class="theme-stroke" stroke-width="1"/>
  <text x="100" y="420" text-anchor="middle" dominant-baseline="middle" font-size="22" class="theme-fill">Contexts</text>

  
  <circle cx="400" cy="420" r="60" fill="none" class="theme-stroke" stroke-width="1"/>
  <text x="400" y="420" text-anchor="middle" dominant-baseline="middle" font-size="22" class="theme-fill">Prompts</text>

  
  <path d="M250 140 L250 190" class="arrow" stroke-width="1.5" marker-end="url(#arrowhead)"/>
  <path d="M135 375 L205 295" class="arrow" stroke-width="1.5" marker-end="url(#arrowhead)"/>
  <path d="M365 375 L295 295" class="arrow" stroke-width="1.5" marker-end="url(#arrowhead)"/>

  
  <line class="dashed-line" x1="205" y1="125" x2="135" y2="375" stroke-width="1.5" stroke-dasharray="4" stroke-linecap="round"/>
  <line class="dashed-line" x1="295" y1="125" x2="365" y2="375" stroke-width="1.5" stroke-dasharray="4" stroke-linecap="round"/>
  
  
  <line class="dashed-line" x1="160" y1="410" x2="340" y2="410" stroke-width="1.5" stroke-dasharray="4" stroke-linecap="round"/>
  <line class="dashed-line" x1="340" y1="430" x2="160" y2="430" stroke-width="1.5" stroke-dasharray="4" stroke-linecap="round"/>
  </svg>
</div>

Most model providers offer limited control over core model behaviors, so the biggest leverage for end users building with models lies in context and prompt. That's where design plays a huge role—both on the backend API structures and the user-facing interfaces.

<h3>A simplified overview of models</h3>

Large Language Models compress massive amounts of internet data into two main parts<sup>[1]</sup>:

- <i>Weights:</i> what the model learned during training.
- <i>Architecture:</i> how it runs during inference.

Once trained, the model predicts the next word based on probabilities shaped by context. But its inner workings are very tough to decipher. It's essentially a black box<sup>2</sup> with no clear dials or knobs. This unpredictability is a major design constraint, forcing us to account for multiple scenarios rather than a single path.

<div class="footnote"><sup>1</sup> Weights and architecture are usually proprietary, though some open models from providers like Meta (LLaMA) or DeepSeek have made both public. As end users of closed source models, we have limited control over the parts of the model that would lead to more significant behavior changes.</div>

<div class="footnote"><sup>2</sup> Interpretability research focuses on understanding how models "think" and reason. <a href="https://www.anthropic.com/research/mapping-mind-language-model">A recent study</a> from Anthropic, for example, found that models often plan several words ahead and generate plausible-sounding responses aimed at agreeing with the user, rather than reasoning through ideas logically.</div>

<h3>Collect and extract contexts</h3>

Professional tasks demand loads of context requirements. LLMs have huge amounts of knowledge but have limited ways to handle context dynamically. They have a context window that caps how much text you can include in each request, plus limited long-term memory. People often solve this with bigger context windows, embeddings<sup>3</sup>, and more precise contexts selections.

Collecting context is harder than it sounds. Tasks we take for granted—like opening a file or searching the web—are non-trivial for LLMs because they weren't trained to <i>see or act</i>. That's why many systems rely on "helpers"—tools, MCP servers, orchestrators, etc. to handle these basic functions. Andrej Karpathy compared the future of LLM systems to <a href="https://www.youtube.com/watch?v=zjkBMFhNj_g&t=30s" target="_blank" >a kernel</a> in an emerging OS, using memory and computational tools to solve problems.

<picture>
  <source srcset="/static/diagrams/llm-context-window-dark.svg" media="(prefers-color-scheme: dark)" />
  <img src="/static/diagrams/llm-context-window.svg" alt="LLM Context Window Architecture" style="max-width: 100%; height: auto; margin: 2rem auto; display: block;" />
</picture>

Even with structured tools, LLMs can skip steps, hallucinate, or make things up as they go. It's a challenge for both engineers and designers. This means designers must push upstream to define the <i>most effective API structure</i> for LLMs, especially if the underlying system is optimized for static UIs. Are we feeding data to the model efficiently? Which tools do we provide for LLMs? How do we help the model understand when and how to use them?

<div class="footnote"><sup>3</sup> There are many practical use cases of <a href="https://platform.openai.com/docs/guides/embeddings" target="_blank">embeddings</a> such as categorization, search, anomaly detection etc. Embeddings models are much much cheaper than regular LLM models.</div>

<h3>Write efficient prompts</h3>

Steering model behavior via prompts is actually much simpler than what you might think<sup>4</sup>. Just be specific, define roles, clarify objectives, and provide examples. Like you're talking to a human with zero context. The style of prompting varies: creative tasks might mean avoiding rigid instructions, while more consistent outputs might require a surplus of examples<sup>5,</sup><sup>6</sup> .

Each model behaves differently. One might fixate on tools over context, another might be great at brainstorming but miss empathy. I like to think of models as <i>quirky child geniuses</i>, and their personalities help me understand their quirks. The best way to improve is to experiment and keep trying.

<picture>
  <source srcset="/static/diagrams/efficient-prompts-dark.svg" media="(prefers-color-scheme: dark)" />
  <img src="/static/diagrams/efficient-prompts.svg" alt="Efficient Prompts" style="max-width: 100%; height: auto; margin: 2rem auto; display: block;" />
</picture>

Prompt-related challenges boil down to two things: increasing the consistency of model behaviors while working with a ton of variables (e.g. context, tools, etc.) and crafting an input-output flow that's easy to use and understand for your users.

<h3>The design challenges, in a nutshell</h3>

Designing with LLMs introduces a unique set of challenges that shift how we think about systems, interfaces, and data:

<i>Create higher value abstractions</i>. The inherent unpredictability of LLMs drives product design toward domain-specific, evaluation-based setups with more predictable behaviors.

<i>Data design matters as much as UI design</i>. Integrate with existing APIs and figure out how to handle long-term, "soaking up" knowledge. It's not just about pretty interfaces, it's also about how data flows and persists.

<i>Manifest the input, reasoning, to output process that provides predictability, transparency, and efficiency</i>. This could mean carefully weaving AI into existing tools in subtle ways—or radically overhauling how we request and process information at its core.

<div class="footnote"><sup>4</sup> Lilian Weng has a concise and to-the-point <a target="_blank" href="https://lilianweng.github.io/posts/2023-03-15-prompt-engineering/">blog post</a> about different prompt engineering techniques.</div>
<div class="footnote"><sup>5</sup> This <a target="_blank" href="https://www.youtube.com/watch?v=T9aRN5JkmL8">video</a> from Anthropic mentioned the importance of being honest, respectful, and transparent, and building ideas through iteration. It mirrors how we communicate effectively with other humans too. <sup>6</sup>Arvid Lunnemark from Anysphere framed the principles of prompt design as how to do <a href="https://www.cursor.com/blog/prompt-design" target="_blank">web design</a>: clear, dyanmic, and adaptable.</div>

<p class="text-center">***</p>

This is for designers new to the space wanting to learn a bit more about the challenges and constraints. It's not a deep dive, but hopefully a solid starting point.

Designing for AI: A Practical Guide


Generative interface demos are everywhere on social media: type a question, watch the interface adapt to help you book a flight or look up a stock. These demos feed our fantasy of a magical, all-knowing black box—personalized and frictionless.

But while AI has transformed how we access information, it hasn’t changed how we process it. Data still needs to be structured into tables and forms so our brains can digest it. Most AI tools follow the same pattern: define, collect, clean, structure, generate. The hardest part isn’t displaying results—it’s deciding what to generate and keeping outputs consistent.

We often want the machine to do more, faster, with no tedious setup.

This mindset keeps resurfacing in how people talk about AI. There’s a persistent gap between the fantasy of agentic systems and the reality of what it takes to make them work. Many cling to the fantasy of the “do-it-all” machines. But behind the curtain, it’s often just cobbled-together grunt work—endless tweaking, testing, workarounds.

This isn’t a cynical take. _Reinventing the familiar matters_. There’s real value in adapting tools to new architectures. But let’s stop pretending any of this happens without human effort. Not yet, anyway.

I keep returning to Marshall McLuhan’s “the medium is the message.” Our tools shape how we see the world. What if AI handled ambiguity or responded to emotion instead of just commands? Could it help us externalize what we can’t name, becoming a thought partner rather than a mere assistant?

Working with AI can feel like a meditation on our own complexity, maybe even a gateway to understanding others. People have tried more ambient, aware experiences, but something’s still missing. Perhaps the real shift lies in transforming the mediums themselves—because it’s the mediums, not the messages, after all.

<p class="text-center">***</p>


There’s a persistent gap between the fantasy of agentic systems and the reality of what it takes to make them work. Many cling to the fantasy of “do-it-all” machines. But behind the curtain, it’s often just cobbled-together grunt work—endless tweaking, testing, workarounds.

The Promise of Generative UIs

<div className="image-container"><img src="https://www.moma.org/media/W1siZiIsIjUyMTYzMiJdLFsicCIsImNvbnZlcnQiLCItcXVhbGl0eSA5MCAtcmVzaXplIDIwMDB4MTQ0MFx1MDAzZSJdXQ.jpg?sha=34aedc40fd6a4ba6" style="width: 100%;" alt="Agnes Martin">
<span className="caption">Agnes Martin. Mountain. 1960</span>
</div>

Over the past year, my work as a software designer has increasingly shifted from drawing interfaces to helping machines to think like humans. It turns out the most difficult part of it all is not acquiring knowledge — modern AI models already have access to vast amounts of it. It’s the ability to <i>disambiguate meaning</i> and <i>manage contexts</i>.

To solve this, we design programs with carefully crafted logic for interpreting tasks. We help models approximate <i>memories</i> and <i>incentives</i>. We write instructions, map connections, and build guardrails to prevent failures.

But the more time I spend thinking about these challenges, the more I started to appreciate how effortlessly human consciousness navigates these complexities and chaos.

<p>...</p>

It reminded me of something I've been reading recently, <a target="_blank" href="https://bookshop.org/p/books/with-my-back-to-the-world-poems-victoria-chang/19994823"><i>With My Back to the World</i></a> by Victoria Chang. Each single poem in the book is named after a painting by <a target="_blank" href="https://www.moma.org/artists/3787">Agnes Martin</a>, an American abstract painter.

Martin described her work as a way to guide viewers into a meditative state, like “walking across an empty beach to gaze at the ocean.” Chang extends this idea further, laying out her words to mimic Martin’s compositions while exploring her personal memories of depression and grief.

Two distinct mediums, themes, and interpretations—seemingly unrelated—yet they coalesce in a way that feels profoundly meaningful.

Unlike machines, we don’t need rules to understand, remember, or reflect — we simply do. Our experiences shape our feelings, and our feelings define who we are as unique individuals. It's our ability to be, not just do.

I want to embrace all those feelings — to experience them fully, to feel them more deeply. Now more than ever.

<p class="text-center">***</p>

We don’t create rules to understand, remember, or reflect — we simply do. Our experiences shape our feelings, and our feelings shape who we are as unique individuals.

The Ability to Be


<div className="image-container"><img src="https://live.staticflickr.com/65535/54280089973_0df93156da_c.jpg" alt="Pizza at June's"/><div className="caption">The special pizza at June's with pancetta, frisée, and chrysanthemum mustard greens</div></div>

<i>It seems to me that our three basic needs, for food and security and love, are so mixed and mingled and entwined that we cannot straightly think of one without the others. So it happens that when I write of hunger I am really writing about love and the hunger for it… and then the warmth and richness and fine reality of hunger satisfied… and it is all one.

-M.F.K. Fisher
</i>

<p>...</p>

So what does food really mean to us?

I felt an <i>urge</i> to write about this. Partly because I ate a pizza that made me cry a few days ago.

The place is called <i>June’s Pizza</i>. They started off as a pandemic pop-up by chef Craig Murli. It shut down three years ago and <a target="_blank" href="https://www.berkeleyside.org/2024/11/01/junes-pizza-west-oakland-return">finally reopened</a> this September in West Oakland, a gritty, industrialized food desert filled with factories, abandoned buildings, empty lots littered with trash, alongside newly built condos. Murli chose to stay rooted here, where <a target="_blank" href="https://www.whatsgooddough.com/blowing-up-then-getting-shut-down-with-craig-from-junes-pizza/">June first got its start</a>.

The space is a converted warehouse with high ceilings, exposed beams. The spray-painted sign displays a simple menu: just two pizzas, one marg and one special. The entire assembly line is in full view once you enter — from the giant cheese wheel in the fridge to containers of densely packed frisée lettuce. It was also, surprisingly well-staffed, a stark contrast to industry trend of <a target="_blank" href="https://www.berkeleyside.org/2024/10/31/jos-modern-thai-turmoil-chef-staff-layoffs">labor cuts and understaffing</a>.

We ordered a white sauce pizza topped with pancetta, frisée, and chrysanthemum mustard greens. It arrived looking seemingly unpolished: uneven edges, a single large charred bubble on one corner, and a little hill of greens on one side. The whole pizza was dusted with pecorino shavings, coating everything.

Start by tossing a few greens onto each slice. The dough was chewy, bouncy, with a subtle texture from the whole wheat. The sauce with pancetta was perfectly seasoned, a hint of green onion adding a touch of sweetness. And the salad mix added that unique variation of texture and bitterness.

<p>...</p>

I grew up eating chrysanthemum mustard greens (tong hao/茼蒿) in my mom’s tofu soups. That medicinal flavor and bitterness was one I had to grow into. But tasting it raw on a pizza for the first time, it felt new yet familiar, and like everything was in perfect harmony.

On the surface, there’s something deeply beautiful about making one single thing, focus on that, and just make it great. But for me, it hits on a deeper level because of where it’s made and how it’s made, the contradictions and imperfections. This pizza felt like a love letter to California, embodying everything it represents—or at least, aspires to represent.

<p class="text-center">...</p>

Food is so much more than just flavors and sustenance. It's a gateway to our collective memories, societal values, and personal identities.

In my hometown, a small coastal city in Northern China, food was simple and unrefined: steamed fish, clams, crabs, and gritty streetside barbecue stands where shirtless men gathered to drink beer and relished charred skewers. It was all about carving out space to enjoy life in ways that were fast, affordable, and plentiful.

A few months ago in Paris, we had a comically French encounter at the classic bistro, <i>Chez Dumonet</i>. A man sat beside us, dining alone. He ordered the duck confit special, devoured it to the bone, and then tackled an enormous slice of millefeuille. When he was done, he asked to meet the chef to thank him. The chef, standing outside with friends, a cigarette in hand, shook the man’s hand. And that’s it, the man walked away.

It struck me how much we can learn about a person and a culture through the simple act of eating. And through observing how people eat, I gain a deeper understanding of them—and of myself too.

<p class="text-center">...</p>

I’ve found inspiration in food writers who capture the profound connections between food, identity, and society. Ruth Reichl wrote about the lack of understanding for grocery costs and <a target="_blank" href="https://ruthreichl.substack.com/p/is-food-political">why food is inherently political</a>. Jonathan Gold captured the vibrant smells and lively fragments of LA, once <a href="https://www.latimes.com/archives/la-xpm-1992-08-23-tm-6990-story.html" target="_blank">writing in the wake of the 1992 riots</a>: <i>"where gunshots and sirens pierce the calm, there are still reasons to stay."</i> Our very own Soleil Ho unpacks the sensitive intersections of race, <a target="_blank" href="https://www.sfchronicle.com/restaurants/article/The-fantasy-and-reality-of-dining-at-Chez-13650410.php">traditions</a>, and politics, and managed to always be unapologetically authentic.

Honestly, I’m not sure what point I’m making. Partly because as I’m writing this, there’s this big pot of my husband’s chicken tortilla soup bubbling on the stove. The aroma feels incredibly familiar and comforting. The weather’s getting cooler, and steam is starting to fog up the kitchen window as I watch him slice limes into perfect little wedges. He’s definitely a precision cutter, contrary to me.

Maybe this is one of those small, ordinary moments I’ll remember years from now. And maybe, when I think of it, I won’t need a reason to feel grateful.

<p class="text-center">***</p>


It seems to me that our three basic needs, for food and security and love, are so mixed and mingled and entwined that we cannot straightly think of one without the others.

Notes on Food


<div className="image-container"><img src="https://live.staticflickr.com/65535/54280483998_27566c4002_c.jpg" alt="A Putia Do Calabrisi, Catania"/><div className="caption">A Putia Do Calabrisi, Catania</div></div>

From the train windows in Sicily, the most common sight was abandoned farms—brick houses without roofs, their surrounding lands overtaken by weeds and cacti. Sicily is grappling with <a target="_blank" href="https://www.theguardian.com/environment/article/2024/aug/19/the-land-is-becoming-desert-drought-pushes-sicilys-farming-heritage-to-the-brink">a climate crisis</a>: drought, rising temperatures, and wildfires. Experts warn that a third of the island could become desert by 2030. These disasters are forcing generations of farmers to abandon their homes.

Amid the swarms of tourists, there’s an undercurrent of counterculture—a frustration evident in the anti-police and anti-mafia graffiti we saw in both Catania and Palermo. For young people here, opportunities are slim beyond tourism.

There's a certain irony here: this sacred land beside the Mediterranean where the Ancient western civilizations flourished and now exists in our collective imagination as a paradise of sun-drenched villas and crystalline waters, is not quite the same when you look inside.

<p>...</p>

Yet Sicily's soul still shines through. The best meal we had was a three-course Sicilian feast at <i>A Putia Do Calabrisi</i> for just ten euros.

It began with olives, caponata, and a tomato salad, followed by Pasta alla Norma and sardines. The server noticed us eyeing the Braciole di Cotenna (pig skin roll-ups) at another table and brought them over for free. They were delicious. The owner sent us off with a big glass of Marsala wine, which we gulped down, leaving us pleasantly tipsy by mid-afternoon.

<p>...</p>

On our last day we visited a small natural wine producer, <a target="_blank" href="https://www.cantinadelmalandrino.it/"><i>Cantina del Malandrino</i></a>, beneath Mount Etna. The owner, Diego, spoke passionately about sustainable farming and the history of winemaking on rich soils of the volcano. We got to learn a lot about the traditional winemaking process and also tasted some samples still in the fermentation process.

As he drove us back to the train station along dark, winding roads, past more abandoned farms, Diego asked what I did for work. When I explained, he suddenly turned and asked my thoughts on artificial intelligence, its threat to knowledge workers, the middle class, and democracy itself. I can’t remember his exact words, but I do remember being left momentarily speechless in the car.

<p class="text-center">***</p>


From the train windows in Sicily, the most common sight was abandoned farms—brick houses without roofs, their surrounding lands overtaken by weeds and cacti. Sicily is grappling with a climate crisis.

Part 3: Sicily, Italy


<div className="image-container"><img src="https://live.staticflickr.com/65535/54278964647_4e103ff7b3_b.jpg" alt="Finn Juhl House"><div className="caption">Finn Juhl House, Charlottenlund</div></div>

On a rainy afternoon, Fred and I were outside a bakery, sharing a pastry and an umbrella. A man pulled up in his car, rolled down the window, and shouted, "Take this," handing us an umbrella. We were a little stunned, and Fred blurted out, “…for free?”

This small act of kindness seemed to embody Copenhagen itself. Everywhere we looked, we found beauty married to practicality, progressive values grounded in tradition, all within a high-trust society. From the unassuming modest house of famous designer <i>Finn Juhl</i>, to <a href="https://en.wikipedia.org/wiki/Amager_Bakke" target="_blank"><i>CopenHill</i></a>, a waste management plant where people hike and ski, to the <i>The Royal Library</i>, where sleek modernity meets carefully restored history.

<p>...</p>

What is this place? Are you telling me <i>Jane Jacobs’s</i> vision of a perfect city actually exists? Why is everyone biking with a smile on their face? Is this the future of humanity?

It’s remarkable to think that just a few decades ago, Copenhagen battled severe industrial pollution, filthy rivers, and choking smog. And yet, it was able to transform itself into one of the most "livable" cities in the world. How is this even possible?

At the entrance of the <a target="_blank" href="https://ddc.dk/"><i>Danish Design Center</i></a>, they play a video about the history of Danish design and its role in building a democratic welfare society. “What is democracy?” it asked. It’s about creating a shared understanding, so we can converse with a common context. And that feels so true here. The culture’s trust seems to come from a deeply ingrained sense of shared rules and values—what’s right and wrong.

<p>...</p>

One night, we stumbled upon a small wine bar tucked between a trans nightclub and a Thai restaurant. It felt like we had wandered into a friend’s living room rather than a bar.A chef from the restaurant had stopped in to unwind, and as it turned out, many performers/workers from the nightclub would come up to drink during their breaks.

Between jokes about sudden loud explosions (possibly gang-related gunfire one guy said) and jabs at overrated Danish design brands (Hay came up), the conversation shifted to Denmark’s lesser-known histories, specifically, Denmark’s colonial history with Greenland.

It’s a familiar and tragic cycle of colonization. The colonizers stripped away the native cultures (one grim example is <a target="_blank" href="https://en.wikipedia.org/wiki/Little_Danes_experiment">the Little Danes experiment</a>). Over time, <a target="_blank" href="https://news.un.org/en/story/2023/02/1133382">the harm spiraled</a>, and the natives became abusers themselves, struggling with drug addiction, poverty, and crimes.

Eerily, after that conversation, we began noticing Inuit people everywhere in Copenhagen. Some begging on the streets. Others, often traveling alone, looking disoriented. Once an elderly Inuit woman burst out of a bus and was nearly hit by a cyclist, but she just kept moving forward.

<p class="text-center">***</p>


On one of those heavy rainy days, Fred and I were outside a bakery, sharing a pastry and an umbrella. A man pulled up in his car, rolled down the window, and shouted, 'Take this'

Part 2: Copenhagen, Denmark


<div className="image-container"><img src="https://live.staticflickr.com/65535/54279846071_19dcc79790_b.jpg" alt="Joan of Arc at the Coronation of Charles VII in Reims Cathedral, by Jean-Auguste-Dominique Ingres, 1854" style="width: 50%; margin: auto;margin-bottom: 12px; height: auto; border-radius: 8px;" /><div className="caption"> Joan of Arc by Jean-Auguste-Dominique Ingres, 1854, in the Louvre, Paris</div></div>

The thing that struck me the most about Paris, as a first time visitor, is the amount of mixed-use buildings and the close integration of where people live and where communities gather.

Everywhere we look, beautiful historic buildings stand atop charming bistros or neighborhood bars. In contrast, <a target="_blank" href="https://www.sfgate.com/food/article/bay-area-michelin-cred-restaurant-lawsuit-19740722.php">a wine bar in our city</a> almost got sued out of existence by a neighbor complaining about noise and crowds. But let’s not dive into all the issues with American cities.

<p>...</p>

Paris just radiates with diversity and vitality, a city with over 2,000 years of uninterrupted human history. The Louvre is a perfect embodiment of this continuity.

I had some visceral moments at the museum. Seeing Titian’s <a target="_blank" href="https://en.wikipedia.org/wiki/Man_with_a_Glove"><i>Portrait of a Man with a Glove</i></a> — its striking portrayal of the fleeting youth encased in this small, fragile wooden frame. I couldn’t help but wonder how many hands had safeguarded it over centuries. Then there was the sculpture <a target="_blank" href="https://www.nytimes.com/interactive/2016/06/27/arts/design/statue-hermaphrodite.html"><i>Sleeping Hermaphroditus</i></a>, depicting the Greek figure embodying both male and female traits. It’s astonishing to think that gender fluidity was a concept that existed over three thousand years ago. In front of Ingres’ <a target="_blank" href="https://collections.louvre.fr/en/ark:/53355/cl010065747"><i>Joan of Arc</i></a>, I noticed a teenage girl in an oversized Stussy shirt carefully reading the descriptions, perhaps reflecting on the challenges of being a young woman today, and pushing against societal norms like Joan once did.

After the visit, I couldn’t stop thinking about how fragile culture truly is. It's so easily hidden, distorted, or lost. These pieces have survived, and I felt so priveleged to be able to see them.

<p>...</p>

I can't talk about Paris without mentioning the food.

Whether it’s a complex traditional dish like the sweetbreads at <a target="_blank" href="https://www.instagram.com/parcelles_paris/?hl=en">Parcelles</a> or the beef bourguignon at <a target="_blank" href="https://www.ourfrenchimpressions.com/josephine---chez-dumonet.html">Joséphine Chez Dumonet</a>, or fresh, creative takes on tradition from unassuming neighborhood restaurants—like the green bean salad at <a target="_blank" href="https://www.instagram.com/cafe_du_coin/?hl=en">Café du Coin</a> or the best razor clams I’ve ever had at <a target="_blank" href="https://www.leverrevole.fr/">Le Verre Volé</a>—it doesn’t matter. It’s always good. The butter is a rich, golden yellow, impossibly creamy, and the wines are universally excellent without breaking the bank.

One thing that struck me was how meals are treated differently. People eat <i>slow</i>. They take their time, and it’s nearly impossible to catch the server’s attention. The concept of rushing through a meal just doesn’t exist.

But the appreication for food is not limited to dining.

People buy fresh produce every day instead of stocking up for the week. They have small fridges in small apartments. There’s very little processed food, which is obvious in all the grocery stores we saw.

Later in our trip, in the South of France, we watched with amusement these two elderly owners of an antique shop in <i>L'Isle-sur-la-Sorgue</i> preparing their lunch. They carefully unpacked fresh cherry tomatoes, grapes, and bread from small containers, setting out silver forks and ceramic plates.

In France, there are things that simply cannot be compromised. And food is undoubtedly one of them.

<p class="text-center">***</p>


Whether it’s a complex traditional dish like the sweetbreads at Parcelles or the beef bourguignon at Joséphine Chez Dumonet, or fresh, creative takes on tradition from unassuming neighborhood restaurants—like the green bean salad at Café du Coin or the best razor clams I’ve ever had at Le Verre Volé—it doesn’t matter. It’s always good.