This is a modified version of an internal post I wrote in 2022 about becoming a Director of Engineering at Facebook/Meta. I’ve edited it to remove the names in the original post as I have not received their permission to post their names publicly. I also removed some references to internal posts and there is no proprietary/internal information. And I never got around to writing the sequel post while at Meta, but I may write it here.

Original Preface

I wrote this post in April 2021 during a no meeting week. I took half a day and wrote 4 pages of notes that were swirling around my head. And then it sat in Google Drive waiting for me to clean it up. Many more no meeting weeks came and re-orgs and fires ate up the time. With many excuses/caveats, I had shown this document to several people and gotten good feedback about its value. What drove me to finally push myself to get through my hesitancy and procrastination was wanting to write the sequel to this post, “What I learned in my first year as a Director”. Let’s hope it doesn’t take another 18 months to write that one.
——————————————————————

I do feel different. I didn’t think I would feel different by being promoted to Director, since at Facebook/Meta, promotions are lagging not leading. So I should have already been doing the job. But with the title, I do feel some pressure to keep earning it. As another Director was giving me advice on their transition to Director, “before you felt like you were punching above your weight”, but now “this is the weight you’re supposed to be punching at and there’s a bit more pressure”. It feels like my impostor syndrome resets once again, “but what if I can’t consistently live up to the expectations”. This is a new bar of expectations and for me personally, now there’s no major initiative like driving the Quest 2 launch to help me “look good”.

Starting on the Path

So how did I get to be a Director? I wanted to write a guide for M2s who are trying to understand what D1 means after being frustrated with not getting it. I was there. In general, I have a very transparent approach to sharing my experiences, so hopefully it helps others get one frame of reference as they traverse similar paths.

After Quest 1 launched almost three years ago, someone told me that I should be promoted for shepherding that landing. At that time, I argued that I actually rolled up my sleeves and did a focused IC half, while maintaining my organization, including a key org leader hire. But I didn’t have the D1 scope, so I was not surprised when I didn’t become a Director after 2019 H1. A few people were surprised I didn’t receive a promotion. My impact was clearly above expectations for an M2, even greatly, but that is different from sustained behavior at the next level.

In 2019 H2, my manager let me know he was retiring and this change greatly frustrated me. I would get a new manager who had to get up to speed on who I was and what I was capable of. But we had just gone through a huge effort with shipping Quest 1. This manager picked me as STL (single-threaded leader) and he saw how I navigated from zero Quest experience to a pretty smooth launch and success. All of this history would be “lost”. After his departure, I was fortunate that a strong engineering lead joined Facebook and became my manager. He did a great job thinking about what I needed to do and the framework for me to think about problems.

Interestingly at this time, I wasn’t sure I wanted to be a Director. Others thought I should be but I was actually undecided. In my over-optimizing head, the elevated expectations were a concern, as well as the lack of mobility to other teams as well as other companies. If I ever wanted a change, I would have to find a Director Level opportunity, where there are fewer opportunities than M2 and the role also meant further away from day-to-day. Believe it or not, I had the same concern when I got promoted to M2. So I told these hesitations to my manager and my skip, I wasn’t sure if I wanted “the jacket” of being a Director. The skip manager and I would talk about trying on the jacket and seeing how it felt. Not everybody wants or should be a Director - it IS a different job.

My skip and I had a few one-on-ones and we talked about my possible paths to become a Director. He stated two key phrases: “Directors make the weather, not report the weather.” and “Directors see around corners”. For the next year plus, I wouldn’t really understand these fully, but now having hit the step function to being a Director I truly understand what they mean.

So advice #1, be clear to yourself and your manager if you want to become a Director. As I am learning now, there’s a lot of work to get someone promoted to Director. A Director at Facebook/Meta is a very high bar. You need buy-in from at least your own VP, if not partner VPs and Directors for across your product group through many iterations on your Director promotion packet. Over multiple halves, you have sustained that Director level impact across many organizations and projects.

So I started on a Director packet for myself but didn’t really dive into it. And by the end of 2019H2, my manager at the time got promoted up in position and I would get a new manager in 2020H1. So this definitely didn’t help my case for a promotion that half, given my hesitancy and all the organizational changes taking up bandwidth to help me understand what I really needed to do. Due to not having a promotion packet ready, it didn’t even get circulated to the various people in the calibration/promotion group. So I wasn’t surprised that I didn’t get promoted for the 2019H2 PSC cycle.

Strangely even with my almost indifference, I was slightly disappointed that it didn’t magically get conveyed to me. Funny that I almost felt entitled to it and didn’t take the promotion seriously. Plenty of people said I was doing the work of a Director and compared me to other existing Directors and said my performance is on par or better. And here is where some madness lies, you can’t base your value by comparing yourself to other people. Some became Directors at other companies and came over as Directors. Others were promoted internally under their own circumstances. There is a frustration when you see the work that you do and go “why am I not a Director”. And without fully understanding what it meant for them and what it means for you, you can find the situation unfair.

What it means being a Director

I looked around at other Directors and observed some implicit realities; Director titles are public, but levels are not. I was told that people will start treating you differently, as you’ve hit that explicit bar. I personally never really cared if someone was a Director or not when I reached out to them. But I have now seen it matter to others as I’ve reached out as a Director. People use the title as short-hand to mean something in the weight of your words or actions. A manager I’ve known for almost my entire time here said “one of my reports said “this really important person made a request and so I was stressed about his request’ and I said ‘oh … that’s Art’”. It takes getting used to and with “great power comes great responsibility”. This helps Directors drive efforts across the org to make big changes to projects/org to “make the weather”. You have to show you can wield this responsibility.

Going for It

There was also an internal pressure inside the organization for me to get promoted as well. I supported managers three levels deep under me. So I was an M2 with an M2 running one of my orgs supporting an M1 supporting an M1. By staying an M2, I’ve compressed the management stack. Growth paths for those managers were limited and I needed to help my team uncompress by getting promoted or moving out of the way.

So in 2020 H1, I got a new manager and I made it clear to him that I wanted to get to Director. I made it a big goal. From his side, my manager told me later that when he took over the role, the situation did not compute. I had an org with clear Director scope, my teams were executing and I had driven an org-wide effort like shipping Quest 1, but I was not a Director. We’ve talked about his observation since - there was a clear business need for a Director to do this job. His framework was that he help me do this job fully as a Director or if I didn’t rise to the needs of the job and he would need to find a Director to fulfill what the organization needed. He was all-in on helping me fully understand what I needed to learn and do.

For 6-8 weeks in 2020H1, I felt really beat up. my manager became an active coach in my gaps to becoming a Director. He would make sure to attend meetings with me and see how I did. Then he would coach me afterwards. It wasn’t that I was doing things wrong, but there were explicit things that I could have done better. Examples: Too deferential/not taking on the responsibility: “Tech Lead and I chatted and they recommend X”. As opposed to “Tech Lead and I discussed and we recommend X”. I was not taking any ownership of the outcome. I was deferring.

In another kick off meeting for a workstream, I was told I was operating at the right altitude by setting the vision, giving the context for the effort, explaining the plan, and appointing those who can drive the pieces. Slowly he backed away and I felt much more comfortable “wearing the jacket” and acting as a Director.

How I ran the ship room was another good example of transitioning to being a Director. For the early days, I made it sound like we were voting on a decision as I was gathering information. For V24, I took on the responsibility of making the decision to cancel or not. I said this is what I know, these are the tradeoffs, does anybody have new information. I think this is more Director level behavior on how we execute at scale. A lot of Director behavior is about how we scale to bigger orgs or bigger impact. There’s more confidence that we’re all moving in the same direction and someone really owns that call.

My manager also noted to me that the verbs became different in the PSC peer feedback. Decided vs Collaborated. “This wouldn’t have happened without Art.” At one level, everybody is key. But I think the nuance is in performing at scale. I think there are lots of opinions and lots of people who are focused on what they are doing. Need someone to rise above it and take in all the input and make a decision, explain the decision and process. After Quest 1, my manager was surprised that I had got Quest 1 shipped and somehow didn’t piss anybody off. Now the question is whether that should have been a non-goal, since pissing someone off can help the organization move forward faster.

Crossing the Finish Line

Luckily I had a manager who was keeping me in the loop on how he was taking me through the promotion process. This is important as now one of my duties as a Director is to get other M2s to Director. The packet writing, the socializing and getting feedback from other Directors on whether they would support it (whether or not they are in the calibration room). The clearly defining where the gaps are, so a lot of the work is done before the actual calibration decision. From M2->D1 is a big jump and you want that decision to be confident.

Over these two years of being a Director I have learned more about these “gaps to director”. What I learned before I got promoted was what was MY personal path to Director, which I’ve shared here. What I’ve learned since then is what the Facebook/Meta bar is for an Engineering Director. This has helped me better explain to others what their path is and part of the sequel post/note.

Final Thanks

Thanks to my manager for his support on this journey. I wouldn't have gotten through it without him. 

(image generated by author with Dall-E 3)

Despite being worth over $30B and CEO of one of the top 6 most valuable companies, Nvidia CEO Jensen Huang responded to a hypothetical question about starting a new company in 2023 with a startling admission: "I wouldn't do it." He explained that building Nvidia had been "a million times harder" than he and his co-founders had anticipated.

"Back then, had we fully grasped the extent of the challenges, the vulnerability, the emotional roller coaster, and the potential for failure, I doubt we would have embarked on this journey. Starting a company is not for the faint of heart."

I've heard similar sentiments from numerous startup founders: "Had I known the journey would be this tough, I might have reconsidered." Now, as I navigate through the early stages of my own startup, I find myself reflecting on these words. There are highs and lows, moments of doubt and bursts of confidence--and I know the emotional intensity is only going to increase.

A year ago, I gained clarity about my top priority in life: ensuring my sons have the freedom to pursue their dream college. I'm fortunate to even have this aspiration, and my ultimate goal is to never let financial constraints hinder their dreams. The projections seem promising, especially given the option I've carved for myself to potentially rejoin a leading tech firm. I've earned the respect in the industry to return as an Engineering Director if need be, ensuring I can meet this personal goal.

With that goal likely secured, I've pondered what truly motivates me and what I wish to achieve in what could be the last decade of my career. The prospect of this startup being my final full-time gig is daunting, yet it's made me realize that I thrive in environments where I can empower others to excel. My knack for navigating complex situations and aligning various elements--people, technology, business, timing, politics--helps create a harmonious and effective workplace. It's not about control; it's about creating a space where I, and others, can do our best work. While there's no guarantee of success, the journey itself promises a wealth of learning and personal growth.

A friend and advisor recently recommended "The Founder's Dilemmas," a book that explores the motivations of startup founders. The author outlines two primary drives: the desire for wealth ("Rich") and the desire for control ("King"). The "Rich" path may require relinquishing some control for the potential of greater financial return, while the "King" path offers more control but potentially less financial reward. I've realized that my position falls somewhere in between. I'm not chasing control for the sake of it, and I'm certainly not motivated solely by financial gain--there are simpler ways to achieve that. Instead, I seek enough influence to ensure a positive and fulfilling work environment, understanding that the journey will be fraught with challenges. But through it all, I am excited to be a part of this endeavor and to learn and grow alongside my team. I think we have a worthwhile product to make. Will it succeed? Maybe, but I’ll learn a lot while trying.

I’m no longer afraid of the worries but excited by the opportunities. I aim to avoid a “You Don't Know What You Got (Till It's Gone)”

I modified a generated art prompt to the following to create the image above: “Depict a scene of reflection and determination, showcasing a person moving forward, away from worries that lead them down to uncertainty, but they are going towards light, optimism, supported by friends and family and opportunity”

Trying Dall-E 3 since I already use Chat-GPT. Not sure if I’m changing to every 4 weeks given startup ramp up, but if I send something out it’ll be Thursday at 11AM PT.

(image created by author using Midjourney)

There are select items I choose based on brand loyalty. When it comes to cookware, All-Clad stands out. I've been using the core set I acquired over two decades ago and remember how its ease of cleaning enhanced my cooking experience. The joy of cooking heightened because cleanup was no longer a chore. Every All-Clad addition, from pizza stones to colanders, has lived up to the standard. I recently gifted a family member a starter set, hoping they'll cherish it as I have. Although All-Clad changed ownership twice, its quality remains consistent.

In an early blog post from the 2000s, I sung praises for brands like All-Clad and TiVo. TiVo, in particular, felt revolutionary at the time. I recall gifting several units to family with lifetime subscriptions. But as competition grew and constraints hampered its progression, TiVo's shine dimmed. It's a stark reminder that superior quality alone doesn't guarantee success; reputation and market presence are equally vital.

KitchenAid mixers are another testament to brand endurance. These robust machines are often handed down generations. While KitchenAid has ventured into other appliances, it's their mixers I'd endorse without hesitation. Their expansion into various kitchen gadgets after being acquired by Whirlpool felt more like capitalizing on the brand name rather than ensuring consistent quality. Fortunately, their core product - the mixer - remains commendable.

Recently, in the midst of my home remodel, I found myself searching for a tool cabinet. Craftsman, once a subsidiary of the behemoth retailer Sears, was a brand I had earmarked for purchase "when I had the funds." For the uninitiated, Sears, before its 2018 bankruptcy, dominated US retail for much of the 80s. Craftsman tools, synonymous with reliability, were every DIY enthusiast's dream. However, my recent research painted a grim picture of their current quality. Despite once offering a lifetime warranty on their tools, Craftsman's commitment waned over time. The brand, now under Black and Decker, has witnessed further constraints, with accountability taking a backseat.

It's intriguing to see how the above brands carved a niche for themselves through sheer quality. Being an ardent advocate for quality, I've championed their products during their heyday. Most of these brands have transitioned ownership, and while some acquisitions led to quality erosion, others have managed to uphold brand value. It's tempting to place blame on corporate acquisition, or challenges like fierce competition and evolving market dynamics. However, my advice, especially to the younger generation, remains consistent: Invest in quality over mere affordability. Though it might sound old-fashioned, there's a world of difference between cheap and value-for-money.

When reflecting on this, I'm drawn to my startup's journey. We are deeply committed to delivering a standout product. This may mean investing more time upfront and tapping into our creative expertise in the short term. However, this dedication will undoubtedly pay dividends in the long run by forging a robust brand. From there, it's essential that we continually validate our customers' trust in both our brand and products.

Prompt generated by Chat-GPT4 for Midjourney: "A tree with roots deep in heritage brands, its branches blossoming with modern gadgets, contrasting the journey from past to present while emphasizing quality throughout."

Bonus image with this generated prompt: "An ethereal bridge connecting iconic brands of yesteryears with the startups of today, each brick symbolizing time-tested quality and forward-thinking ambition."

I am a huge Lego fan and originally planned on putting Lego as one of the brands in the post, but felt it was a bit of a tangent to the main topic (and they’ve stayed independent). But I didn’t chose this image because there isn’t a transition with the two sides of the bridge.

(image generated by Author using Midjourney)

This is a spiritual follow up to the post I wrote six months ago about prioritization and working on urgent but not important items instead of important but not urgent:

I've gotten into the habit of not being on top of writing these posts in a timely manner. Once again, I'm writing this on the Thursday morning, with the 11AM posting deadline looming. And even looking at the last 4 posts, two were delayed due to moving/remodel and children sports ramping up on the weekend, one was a post that took two weeks to write, and this one was last minute. I haven't prioritized writing these posts because my mental energy hasn't been inspired. There is no muse inspiring me "you must write about this topic".

And I think I'm at the point where I need to adjust my process. Possibilities include, writing less frequently (makes sense with startup ramping up) or changing my creative process. I could adjust it to once every two weeks or once every month. There's also, raising the priority (importance and urgency) on drafting a topic on Sunday. The other thing I'm managing isn't necessarily time but also mental energy. I definitely want to continue to write, to put in public some of my thoughts on lessons learned and perspectives. But I feel like I need to have something worthy of your time and of my own future time to read.

I'll start with a bit more structure on coming up with ideas earlier, fleshing them out, and not just procrastinating until the last day. Was this all a temporary blip in my process due to the personal life disruption due to move back home and the startup ramping up? Where do the adjustments need to be made? Let me try a few things and report back.

Image prompt: “An abstract balance scale, one side weighted with sheets of paper and a quill, the other with a vibrant muse being overshadowed by an encroaching clock.”

(Image created by Author using Midjourney)

I'd often heard how declarative UI frameworks, like React and SwiftUI, stand in contrast to imperative ones. To put it simply: in a declarative approach, you outline your desired result and let the framework do the rendering, whereas in an imperative method, you lay out step-by-step instructions to create the UI. While the latter grants more control, it often demands more detailed instructions.

When it came time to develop a UI for our startup's prototyping, I was tempted to lean into SwiftUI, considering our aspirations for a mobile app. Yet, I settled on React. The reason? A plethora of learning resources, greater longevity, and quite frankly, better Chat-GPT compatibility. Diving into YouTube tutorials, I got the hang of the syntax and followed along with the exercises. Yet, something didn't resonate. I could pen down React components, and conceive pieces of a UI using them, but when it came to crafting an app from scratch, there was a disconnect. While YouTube tutorials make complex topics more digestible, I felt they sometimes skimped on the larger picture.

A week into this exploration, I stumbled upon some React-based game engines on GitHub. These provided more than mere UI component blueprints; they offered the missing link. Initially, I pondered if the week spent on tutorials was redundant. Should I have started here? However, I soon appreciated that those tutorials grounded me in the basics, enabling me to decode and grasp someone else's code.

There were moments of doubt along the way--wondering if my years of imperative programming had built walls against understanding newer paradigms. But with React and other declarative frameworks gaining traction, I held onto the hope that things would eventually click. And they did! However, I realized I'd delved too deep into nuances without the anchor of real-world applications. Context, as I learned, is crucial.

Having spent my career in the realm of imperative programming, I relished the precision it offered. It's intriguing how declarative UIs essentially serve as a narrative, directing the imperative "engine" on rendering specifics. At its core, computers process code, after all. Declarative design abstracts the nitty-gritty, simplifying the process--provided you play by its rules. I'm optimistic about prototyping with declarative, but I wonder if it can cater to our need for finesse and uniqueness when we're ready to add some flair. Many are drawn to the allure of 'easy mode', but it often comes at the cost of customization. In the startup world, a standout UI can make all the difference. What will the final product be written in, we'll see when the time comes.

Chat-GPT4 created his prompt for the image: “A puzzle in the making: individual, well-defined pieces (representing imperative practices) gradually blending and merging into a fluid, interconnected tapestry (declarative). Ethereal code threads weave throughout, hinting at the undercurrent of learning and adaptation.”

(meme generated on Imgflip)

I recently penned a piece about learning React. While I'm sure Chat-GPT could polish it up, I felt it lacked depth and substance. Since moving back into our home, I've been jolted awake at 4AM for the past couple of nights, besieged by an extensive to-do list and a sea of boxes.

Surprisingly, I only managed to set up my computer late into the second day post-move, which is quite uncharacteristic of me. Instead of rushing out a post that doesn't meet my standards, I've decided to let the ideas simmer a bit. Expect a more thoughtful piece next week.

On the home front, after seven months of remodeling alongside a phenomenal contractor and designer, it's surreal to realize this is my living space. Everything feels simultaneously familiar and wonderfully new. Rest assured, a post on the craftsmanship that went into this transformation is in the pipeline.

See you all next week. Your patience is much appreciated.

(image created by Author with Midjourney)

Thursday crept up on me amidst the hustle and bustle of finalizing our home remodel and moving back in. It's quite the surprise for everyone that our contractor is ahead of schedule. We're looking at just shy of seven months, whereas the initial estimate was 7-9 months. I was aiming for Thanksgiving and thinking Halloween would be a treat, but getting settled by Labor Day? What a luxury! Both our contractor and designer have been stellar – meticulous with every detail. When I inquired about our ahead-of-schedule progress, our designer mentioned it's due to our swift decision-making and commitment to those choices. And trust me, there were countless decisions at every turn. The best part about moving back? Every little detail in our home was handpicked by us.

Apologies for the brief update today; I only remembered it was post day this morning. Next week might be another concise entry. I'm mulling over a topic on the highs and lows of learning React and my personal strategy for mastering new tech (it's more about effectiveness than efficiency).

(image generated by Author with Midjourney)

Nearly three decades ago, at my first job out of college, a colleague dubbed me the most sentimental person they knew. As a young adult, I was surprised to discover that my proclivity for cherishing memories wasn't universal. With the wisdom of age, I now see it as one of my distinctive qualities. That first job was a career highlight. Reflecting on that team, I realize I didn't fully appreciate the autonomy, leadership, product, and sheer joy of those days. Oh, how I wish I could encapsulate that experience and relive it. As time passed, that incredible team disbanded and branched off, accomplishing remarkable feats independently.

That's the thing about great teams. Over time, dynamics shift. Personal lives evolve, skills grow, members come and go, and the character of relationships transform. So, for those fortunate to be part of a fantastic team, treasure it. That unique blend of personalities and talents might not come together the same way again, even if the roster remains unchanged. Every project, every year brings fresh challenges and altered perspectives.

Being a die-hard sports fan, I see parallels in the ever-changing dynamics of teams. Players chase higher salaries, while management shuffles the roster to optimize performance. The sports world operates on a seasonal rhythm, with fixed contracts and variable performance bonuses. A team's chemistry this season won't be identical next season, influenced by skill development, competition, and other evolving factors. Every season is a unique snapshot in time, meant to be savored by players and fans alike.

This sentiment of fleeting moments is perhaps most palpable in parenthood. Friends sending their kids off to college grapple with the reality of an empty nest. While those days of bustling homes might return temporarily, the dynamic forever changes. From being the primary caregiver, the transition to watching them make independent life choices is profound.

On a more personal note, a friend and former colleague recently mourned the loss of their cherished pet, spanning over a decade. I can still feel the profound sorrow from the days my previous dogs departed. With my current loyal companion, the inevitability of loss lingers in the back of my mind. All I aim for is to offer him the best life possible. My previous dog, who was with me through significant life changes, left an indelible mark on my heart. When he passed, my brother advised me to jot down cherished memories. On every angel day (the anniversary of his passing), I revisit that note. While the initial pain has subsided, that note rekindles precious details that would have otherwise faded. The details captured in those lines help me relive and treasure the moments we shared.

Savor the moments you have today. Life is too short.

I had to generate 9 different prompts to finally get some artwork that I thought best fit this article. ChatGPT was a bit too literal with sports, dog on some of the prompts. The prompt that generated the artwork above is: “Depict a tapestry of time, where cherished moments linger like golden threads amidst the ever-shifting fabric of relationships and life transitions.”

(image generated by author on Midjourney)

The topic of Return to Office (RTO) is all the rage now, particularly among the tech giants. There's this tug-of-war between carrots and sticks to bring folks back to the physical workspaces. Personally, I thrive when working alongside my colleagues in the office. I remember heading back as soon as it was permissible during the Covid era. The camaraderie and collaboration that comes from being in a shared workspace drive my enthusiasm for work. Even at Facebook, sprawled over numerous buildings and floors, I'd much prefer a stroll to a colleague's desk over a quick ping on Messenger. Direct interactions simply had a richness that digital communication sometimes missed. There's a plethora of micro-interactions in the office that foster empathy and mutual understanding. Particularly in my role, overseeing the VR program, these interactions were invaluable for weighing pros and cons, enabling me to make the best decisions possible. Collaborating with sharp, inventive minds necessitates this close-knit interaction--it's about laying a foundation of genuine listening rather than seeing every discussion as a contest.

This past weekend, the merits of in-person interaction became evident during a whiteboard brainstorming session. It's incomparable, the ease of picking up a pen and adjusting a colleague's note to make a point, all in real-time. Compare this to a Zoom call later in the week with a former team: the occasional awkward pause, the "Oh, sorry, you go first" dance. Existing relationships can navigate this digital realm with understanding, but new ones? They can easily misstep, treading the line between politeness and frustration over perceived slights. You really need face time, in my opinion. There’s always iteration and back and forth to get to the best solution and reducing the lag is critical.

While I don't advocate for a full week in the office, regular coordinated in-office sessions are pivotal for groundbreaking work. Apple's insistence on office presence--given their emphasis on interdisciplinary integration--makes complete sense to me. Yes, working from home offers undisturbed focus, ideal for linear tasks. However, groundbreaking projects--those "zero to one" ventures full of unpredictabilities--demand cohesive teamwork and immediate feedback.

From my vantage point in the gaming industry, the "hey, check this out!" moments, seeking instant feedback or kindling team spirit, are invaluable. It's arguably the most efficient approach, in my book. I recognize the challenges: taxing commutes, living costs, and the lifestyle adjustments from the past few years. However, I sense a shift back towards valuing in-person interactions. Our startup is considering a blended approach, emphasizing regular face-to-face meetups. Locally, we're eyeing a schedule of 2-3 in-office days per week. I've even repurposed my garage into the quintessential Silicon Valley startup hub, complete with a whiteboard--a space for birthing revolutionary ideas. From there, it's all about executing in our respective optimal settings. We'll report how it goes.

Above art generated by prompt modified from one generated by ChatGPT-4: “Interwoven tapestry of a Silicon Valley warehouse start-up with whiteboards and explosion of ideas and collaboration with a monitor showing another person on Zoom”

Bonus Images

I couldn’t get the car out of the ‘garage’ for the prompt even adding the “(with no car)”: “Interwoven tapestry of a Silicon Valley garage (with no car) start-up juxtaposed against a virtual Zoom grid, with threads of creativity, trust, and innovation binding them.

Another fun one but I didn’t feel the vibe (nor all the suits!): “Dynamic dance of office desks and virtual meeting screens amidst a backdrop of Big Tech skyscrapers, all under the overarching aura of collaboration and change”

(image generated by Author on Midjourney)

"He's too young to be Vice President," my co-founder informed me, quoting someone he'd met. They'd seen my title of "Project Leader, VP" on my business card during a meeting with partners. We were just a startup with a team that could be counted on one hand. I was taken aback by this comment as I hadn't given much thought to my title when we had the cards made up. This was my second job out of college, and my co-founder had been the president/founder at my first job, so I wasn't an unknown quantity to him. However, to this person, who had experience working in larger companies, the title held significance.

I find it somewhat ironic that job titles often seem to matter most to those who need them to do their job. Here are some other stories about titles:

Earlier in my career, a designer I supported asked to become the lead designer after only a few months on the job. I granted the request without much thought, only to see the designer act as though the title gave them more authority, despite there being no other designers on the team. In my opinion, if you require the title to lead or be listened to, then you're not ready yet. In almost all of my experiences in development, great products are done by teams that listen to the best ideas or arguments, not to titles. One of my biggest warning signs for leaving a job/company is hearing "I told you so" due to their title.

Facebook/Meta's promotion policy operates on a principle of lagging rather than leading. In other words, you need to prove your competence at the next level before the promotion is granted. I penned a detailed internal post about my journey to becoming a Director. A few people said my post was the "what's not on the official wiki about how to get promoted". I discussed the struggles I went through to understand what it truly meant to be a Facebook Director, a role I did not fully comprehend until I received guidance from an excellent manager. At a big tech company, I finally understood the bar of what this title means. For better or worse, it's also shorthand for any future employer I may have.

My point of view at work is that titles are very useful as a starting reference point but the person's actions/words need to back it up. It's good shorthand when you talk to another Director at Facebook because you know the likely scrutiny they received during their review process and how they should be thinking about things. But things aren't always fair or consistent, whether it's a promotion or a hire that didn't match my expectations. It can also go the other way, as many people were mentioned at review time whom I hadn't seen before, and a bunch of us said, "I didn't know that person was at that level," as Facebook/Meta doesn't share out levels.

One must avoid a sense of entitlement that titles or tenure can bring. An early member of Microsoft once noted that some peers felt they were always right because of their wealth, which they attributed to winning "the job lottery," as they had multiple options and just happened to have chosen Microsoft. There were people who wouldn’t push back on me because I was “their Director”, which I found disconcerting as it made me a title and not a person.

In the end, titles serve as important milestones and shorthand, especially in large companies where big teams must cooperate. They will never be perfect or entirely fair, and overemphasizing them can lead to misunderstanding and disappointment. They are only one way a company values you, but remember, there are many other ways you can be both valued and undervalued. Pay attention to those as well.

Note: I'll be taking next week off from writing a newsletter due to some single parenting duties. Be back on August 17, 2023.

Art above created with prompt by ChatGPT4 into Midjourney: “Visualize a journey of a young entrepreneur, a combination of corporate ladders, and an abstract representation of a big tech company. Show the contrasts and intersections between youthful energy and established structures.”

(image generated by Author with Midjourney)

The question I often field is, "how's the startup?" Starting from scratch is a rollercoaster of highs and lows. Having embarked on this journey four times before, the variance isn't surprising, but it doesn't necessarily make it easier. This time around, the challenge is even more thrilling, as I'm diving headfirst into a technical domain I had little experience in: Deep Learning/AI. Over the past two months, I've immersed myself, learning Python, LLM, tokenizers, embeddings, git, and cloud deployments, to name a few. And still, there's a mountain of knowledge to become fluent in to create a truly magical product.

Gaining momentum as a company has been challenging. My co-founder, Constantin, is in a European time zone for part of our journey, and we're also waiting for another co-founder to join us. Despite the distance, our daily touchpoint has been essential to share our triumphs and tribulations, supporting each other through frustrations and disappointments. I'm excited for the time when we can collaborate in earnest.

In the midst of all this, I must admit that seeing Meta's stock price soar straight upwards since my departure has been a minor distraction. Especially during the hard weeks, when you hit walls or make little progress, the financial stability of a salaried job seems attractive. In a large organization, your productivity level might fluctuate, but the income remains constant, and the company's momentum carries you along. It is still the right decision for me and that inertia cuts both ways. However, at a startup's beginning, you are completely the tone setter and that’s both the curse and prize.

To keep pace with this whirlwind, I've been absorbing as much information as I can. Tutorials, open source projects, and YouTube videos all depict the "golden path," seemingly perfect solutions that, when scrutinized, often crumble. I'm gradually digging into source code, not merely to utilize it, but to understand it, building a sturdy foundation.

And sometimes, like last week's introduction of the "open source" Llama2 to the AI scene, everything resets. Initially, I felt like a novice attempting to harness its power. Now, with more comprehensive resources becoming available, progress is visible. Given Llama2's significance, I believe we'll be sticking with it for some time.

Aside from software, I also had a valuable realization last week. After experiencing issues with cloud servers, I decided to get the best consumer-grade video card for AI inference. The increased reliability and processing power has substantially accelerated my learning pace. I wish I had made this upgrade sooner, as it has helped me make and learn from mistakes more quickly. I've also discovered that I can connect my MacBook Air to my Linux machine, facilitating a more comfortable and familiar workflow. This is all contributing to my velocity.

Today, I'm almost done getting Llama2 to parse my 34(!) posts here, teaching it to write like me. It's not generating posts for me... yet.

Midjourney art prompt generated by Chat-GPT4: "A high-tech startup spaceship navigating through a galaxy of deep learning symbols and code snippets, representing Python, LLM, tokenizers, and embedding. It's fuelled by lines of code, with two shadowy figures (the founders) at the helm. In the background, a bright star (symbolizing a successful outcome) is rising on the horizon." One trick I started doing is asking for 3 art prompts and trying each of them.

(image generated by author with Midjourney)

I'm genuinely thrilled to delve deep into the topic of AI in this article. The acceleration of development in this field, particularly since the launch of ChatGPT on November 30, 2022, is truly humbling. My co-founder and I had already decided to venture into the AI/ML realm even before this significant milestone. Since then, AI has taken the tech world by storm and has become a hot topic across many family holiday discussions. Views range from those who foresee an AI takeover or human redundancy to skeptics who dismiss it as a fleeting trend akin to recent cryptotechnology. Today, I'll share my insights on AI, derived from my full-time exploration of this technology.

If you're a regular reader, you won't be surprised by my pragmatic stance on AI. I perceive it as both more and less remarkable than commonly anticipated. Up close, it's clear the technology still has a considerable journey before it matches the hype generated by both enthusiasts and alarmists. However, I view AI as an invaluable productivity tool that can alleviate the tedium of routine tasks. I've personally found Github Copilot's programming assistance incredibly helpful in managing boilerplate code or filling in forgotten syntax. It effectively does what Stack Overflow once did for me: sourcing solutions to common problems. Generative AI has substantially streamlined the task of finding and understanding existing information. The breakthrough of ChatGPT lies in its ability to comprehend and communicate in natural language, making it a more user-friendly interface to software and computers.

An interesting study conducted earlier this year used ChatGPT-3.5 to emulate regular tasks. It concluded that ChatGPT users were more productive, efficient, and enjoyed their tasks more. Particularly, individuals with weaker skills gained the most from the tool, suggesting AI's potential role in reducing productivity inequality. However, the utility of current AI doesn't necessarily transform users into superhumans, but rather offers significant assistance when effectively utilized. It also allows people to go into new domains much easier, which I think people are underestimating the impact to society.

During a conversation with a good friend about Meta's current performance summary cycle (PSC), they expressed discontent over people using generative AI to trim their summaries to the mandatory 300-word limit. I queried the problem with that approach and the reason for the word limit. The rationale, to save time and keep the summaries concise, seemed to justify the use of an AI tool to achieve that goal. Especially for those who aren't adept at English or tend to be verbose, isn't this tool a leveller? While individuals are still entirely accountable for their summary content, doesn't a tool that enhances efficiency benefit everyone?

I've personally utilized AI to refine these newsletters/posts. After jotting down my initial thoughts and ideas, Chat-GPT tidies it up, saving me time. Although it streamlines the process, I ensure the content remains authentically me and engaging for you, the reader. Recently, I assisted a friend applying for a University position by crafting a recommendation letter. I wrote in my typical style, and the AI tool was able to transform it into a more suitable language for that context. Despite this, I've noticed it can occasionally strip away some style or tone to achieve clarity, which can render the content somewhat bland.

So, AI is an efficient tool for communicating existing ideas and knowledge, which aligns with its training objectives. The last few decades have seen mass-media fantasies about AI (and the Metaverse), leading to extensive speculation about its potential or consequences. I recently fielded a question about whether our AI startup would evolve into the equivalent of Skynet from Terminator. I laughed it off, explaining how these machines work. But who knows what other researchers are cooking up in this space? What's the next giant leap for AI? It's fascinating how the effective user interface of ChatGPT atop the existing Chat-GPT3 model captured public imagination and generated hype. I'm genuinely thrilled about this field's prospects and our contribution to its evolution. Stay tuned.

I think this is the last time, I’ll say that the post was edited by Chat-GPT4.

Image prompt generated by Chat-GPT: "Please generate an image of a vibrant garden, full of diverse plants and flowers, symbolizing the wide range of AI possibilities. Include a gardener in the scene, diligently pruning and nurturing the plants, depicting the ongoing efforts to improve AI. Lastly, incorporate a futuristic looking tool in the gardener's hand, representing AI's role as a productivity enhancer."

Bonus image by Midjourney:

(image generated by author with Midjourney)

(created by author with Midjourney)

Craftsmanship. Polish. These are terms often tossed around, but what do they really mean? Let's explore this by looking at two areas where I've personally sought to hone these qualities: mixology and software development.

Over the last few years, I've developed a hobby of making the Old Fashioned cocktail. Whenever I'm at a restaurant, I find myself comparing their concoction to mine. A successful Old Fashioned goes beyond just taste—it involves a blend of visual appeal, texture, temperature, and even the glass it's served in. The craft of cocktail-making demands a deep understanding of ingredients, the role of water and ice, and the science of flavor and experience. To enhance my skills, I've experimented by altering just one ingredient or technique each time I make an Old Fashioned. This helps me understand the variables at play when someone requests a specific flavor profile. To deepen my knowledge, I delved into the ‘Cocktail Codex’, a gift from a colleague and Instagram star mixologist, Parth Vasa, which breaks down the six foundational cocktails and their derivations. Similarly, Scott Moore's book ‘Liquid Intelligence’ elucidates the science behind cocktail-making. By grasping not just the 'what', but the 'why', I can experiment with and elevate my creations.

In software development, I began my career working on 'AAA games', which demanded a high level of quality and production. The term that resonated with me was 'polish'—creating a user experience that feels just right, beyond mere functionality. Achieving this requires an understanding of user friction points, and a commitment to iterative refinement. It's rare to nail it in one go; more often, it's about getting a basic version running, and then investing time in the 'polish', whether that's refining the code or tweaking the output to perfection. An example from my personal projects is my work on a discord bot; I spent considerable time determining the ideal formatting to ensure the bot's daily updates could be quickly and easily understood.

As I delve into the world of AI, I've found that it's often presented with a veneer of hype and 'clickbait' promises that don't hold up in practice. AI can achieve remarkable things within narrow parameters, but when you venture off the beaten path, things get complex rapidly. There's a clear trend of users seeking easy abstractions, often at the cost of a deep understanding of the underlying mechanics.

My own approach leans towards ensuring things work efficiently and beautifully, in that order. I believe in harnessing the quirks of AI—like the occasional nonsensical answer from ChatGPT—as a strength, rather than a drawback. To achieve this, I strive to understand the workings of AI deeply so I can fine-tune and adapt its processes. This is crucial in the emerging field of Large Language Models (LLM), where the core 'black box' requires the perfect 'incantation' to produce the desired result. Understanding the surrounding environment of the 'black box' is critical to control its output. Without this knowledge, you risk ending up with something bland or a chaotic mess. Be it a cocktail or an AI-powered product, the aim is to craft an experience that feels magical and gratifying.

I was in a rush given vacation week and almost didn’t post. I drafted a few things together but I was amazed at how well ChatGPT actually cleaned it up to something a bit more cohesive with a few changes at the start and end. So yes, today’s update was edited by ChatGPT and allowed us to have this post today.

Image above created with this prompt: "Generate an abstract representation of the process of crafting an Old Fashioned cocktail and refining a piece of software. The image should symbolize the journey from raw elements to a polished product. Incorporate elements like the distinct ingredients of a cocktail, flowing lines that indicate iterative refinement, and symbolisms of coding or software development. The image should evoke a sense of blending science and art, and convey the beauty of refinement and craftsmanship. The color palette should be warm and inviting, to reflect the experience of enjoying a well-made cocktail or a smoothly running software."

Bonus Art: I almost used this instead of the above.

(created by author with Midjourney)

Created with prompt: "Create a cross section image of an architectural diagram of a building and the ground it stands on, starting from its deep, intricate foundation at the bottom and transforming into an impressive, architecturally striking structure at the top. Let the hidden view of the inside of the building and everything under ground demonstrate the importance of a solid, well-understood foundation and the meticulous craftsmanship involved in turning a blueprint into a masterpiece."

(Imagine generated by Midjourney)

You know what they say: success is about positioning yourself to get lucky. As I reflect on my career's wild ride and the birth of my fifth company, I couldn't agree more. Yes, there's a fair share of innate skill and sweat equity involved, but there were also moments where I found myself in the right place at the right time.

In previous posts, I've shared about winning the career lottery a few times. My first jackpot was a summer internship with Looking Glass Technologies, working on the project System Shock. I got "lucky" when this game turned into a cult classic, even scoring a recent modern remake. But, behind that luck was my relentless persistence, visiting the MIT career center weekly, signing up for countless interviews, and flooding the top 20 game companies with resumes. Warren Spector found it amusing that Origin Systems (the eventual publisher of System Shock) sent me a generic rejection letter.

After my Looking Glass interview, I was told it could take weeks to process internships. Surprisingly, they invited me for an on-site interview within three days. From there, my passion and diligence transformed the internship into a part-time role into a full-time one, and eventually a leadership role. Sure, I had summer offers from Apple and Microsoft, but I chose to pursue my gaming passion. Who knows? Those could've been lottery tickets too, considering where those companies stand now.

My classic Asian father deserves credit for instilling in me a relentless work ethic. Yes, it's a cliché, but it was that grit that carried me to MIT and opened the door to Looking Glass. There, I met co-founder Edward Lerner, which led us to start a company together - Multitude. From there, I joined Valve, thanks to a recommendation from my mentor and friend from Looking Glass, Doug Church, to Gabe Newell. Every opportunity had a thread linked to a previous one, each role a piece of a bigger puzzle. My entry into Facebook started with a post about shutting down my startup and job hunting, and three Oculus friends told me I needed to join.

The leap to Facebook presented another lottery - joining as a Manager Level 1, demonstrating my ability and initiative to Level 2, and then running for Director. I was also handed the opportunity to lead the Quest 1 program - I was in the right place at the right time, again. Initially resistant because of my love for the PC/Rift program, I decided to "drink from the Quest firehose," as my manager, Bruno Silva, put it. My hard work and diligence "forced" me into the position. Fast forward three years, I directed the launch of three major VR headsets, some true career highlights.

Success may seem 'instantaneous' to onlookers, but it rarely is. It's an overnight phenomenon decades in the making. I've closed startups when success was elusive and walked away when there was no clear path forward, but each experience made me grow and fortified my resilience. Fortunately, during my stint at Facebook/Oculus, I met my CEO-founders for my new venture, two remarkable individuals who brought out the best in me. Now, as I embark on my newest venture, I can't wait to see how this journey will unfold, promising yet another 'overnight success' built on decades of perseverance and growth.

Text above edited by Chat-GPT4. It also generated the art prompt for the image above: “Create an artwork featuring a series of stepping stones across a body of water, each stone representing different phases of a career. On the far bank, there's a new day breaking, symbolizing an 'overnight success'. Show the ripple effects on the water, illustrating the impact and growth from each step.” I tried creating a more iconic version but those didn’t come out well.

Following the excitement of the Threads launch, I am on there as mrminman and will likely start posting a link to these updates there. This is my 31st post and it feels like I’ve got the momentum going.

(image created by Midjourney)

At Facebook, I was doubly lucky to team up with two fantastic product partners. First up was Brandon Dillon, who nudged me to join Facebook/Oculus when I was in job hunt mode. Together, we rolled out a metaverse product with a tight-knit but super talented team. We had this routine where we'd jot down our priorities for the coming month, then swap notes. Even though I knew my list was more of a suggestion, we'd get into deep chats about why I ranked things the way I did. In a meeting one time, Brandon flipped a full 180 on me, and my brain froze up for a day while I tried to recalculate how we were going to move forward.

I was fortunate again with Constantin Koumouzelis, Brandon's successor. He and I joined forces on shipping Quest 2 and navigating the VR Platform's core software. As VR began to bloom inside Facebook/Meta, our to-do list exploded. Not only did Constantin and I had to discuss the what and why to get stuff done but also the who. We had to steer the ship through the crowded waters of the organization, and we needed to be on the same page a lot.

Both Brandon and Constantin shared a lot in common – programming backgrounds, experience in the games industry, and a whole lot of respect for different perspectives. I wasn't shy about chipping in with product opinions or them with technical opinions. We meshed our expertise, had tons of talks, and made sure we got where each other was coming from.

For me, that's what partnership is all about. I couldn't exactly step into their shoes, but I could stand in for them when needed – and they could do the same for me. I remember another person who thought we had a great "partnership", but it felt more like they were just into dividing and conquering tasks. True partnership, for me, is more than just splitting the work - it's about being able to represent each other.

Partnerships aren't limited to your peers. I deeply considered Justin Rogers and Gerhard Schneider, who I supported, as invaluable partners. They were both technically savvier than I, but our relationship, just like with my product partners, was founded on deep respect and honesty. I often found myself admitting, "I don't know" to them or suggesting, "Let's figure out this problem together." I made it a point to learn as much as possible from both of them. This honest dialogue paved the way for us to be brutally frank with each other, reinforcing our partnership.

Fast-forward to now, I'm super excited to be hitting the ground running on a new adventure with Constantin, my recent partner in crime. We're both full-on back in the game, working hands-on at our startup in the AI space. With so many tools and APIs, both commercial and open source, it feels like we're kids in a candy store. We love building again and have spent many late nights "just to get it right". But even amidst the coding and planning, we're always checking in on each other. It's about holding each other accountable but also remembering that we're human. This kind of partnership brings a level of trust and respect that makes the wild ride of a startup a whole lot less daunting.

Text editing by ChatGPT-4. I used the prompt ChatGPT told me last week to use in the future, and it still stripped away too much personality. I asked it not to and got a much better answer. The text above still has me inserting back some flavor.

And here’s ChatGPT’s 2nd prompt for the image above: “Create a piece depicting two figures navigating a labyrinth together, illuminated by the glow of a shared compass. This signifies the partnership, shared journey, and the light of innovation." The first prompt was 3x longer which seems to confuse Midjourney. I also had to re-run it 3 times to get one image that worked well.

(Image courtesy of: UCLA’s Office for Equity, Diversity, and Inclusion)

Last week, a buzz about an upcoming long weekend was going around. That's when I realized, "Wait, it's Juneteenth!" This marks just the third year of Juneteenth being recognized as a federal holiday. After George Floyd's tragic incident, Facebook initiated Juneteenth programming which was deeply informative to me. I'd attended a few diversity discussions before, but the speakers on that day, particularly Professor Michael Dyson, left a profound impact on me. They prompted me to contemplate the unique challenges black people face in our country.

Seeing our country elect and re-elect a black president shows progress, and having two national holidays (MLK Jr. Day and Juneteenth) acknowledging black people's struggles is notable. This reality wasn't always clear to me, but I made a conscious effort to educate myself. In my daily posts at Facebook, I would often include a sentence or two about #blacklivesmatter, further deepening my understanding.

I believe people need to reflect on this country's unsettling history at their own pace. It's undeniably hard to grapple with the long-lasting implications of what this country has done. I don't aim to spark a woke vs anti-woke debate or pose as an expert on this topic. But I do hope that you might question why Juneteenth is a national holiday for our country. I'm still learning about this topic myself, keeping an open mind about its impact on people mentally, culturally, and financially.

This topic surfaced recently when my co-founder and I discussed whether we should observe Juneteenth as a holiday at our new company. We quickly agreed and wanted to establish a principle that our company respects this holiday and what it signifies for people, ourselves, and our future customers and employees. Our Facebook/Meta tenure was vastly different, where most administrative decisions were made for us. Now, as leaders of our own company, we are responsible for these decisions. It's an exciting time, with a blank slate to shape our company's future.

We are in the process of incorporating our company this week, and hopefully, I'll be able to reveal our official name in next week's post. I'm also eager to share my thoughts on my first few weeks in the AI/Machine Learning/Deep Learning field. As I've mentioned before, I'm programming again full-time for the first time in over six years. I am eagerly diving into a world of REST APIs and many open-source tools and libraries, helping me get up to speed. But more on that in next week's post.

Programming, to me, feels like magic. I'm excited to build and even more excited to share our creations with you all, once we decide what that is.

Note: I wrote this draft and debated posting on Monday. But for reasons of quality and consistency, I stuck to the regular process of iteration. Hopefully, this post can serve as a simple reminder of what Juneteenth represents and allow people to process it in their own way.

Besides striving to be more open-minded and learning about the historical challenges of being black in today's society, after George Floyd’s murder, I started a recurring donation to the Equal Justice Initiative, an organization rated with a perfect score by Charity Navigator.

(Image created by Midjourney)

In the past, I've mentioned the importance of foresight and rapid prototyping in successful game development. Building a great game often involves the parallel process of crafting content and constructing the technology that powers it. This necessitates a team or leadership with the capacity to envision the future of the game, predicting the impact of various technical or design elements before they're implemented. Those who only trust what's directly in front of them often feel the need to build a concept out entirely before validating it. If it doesn't meet their expectations, they scrap it. However, this approach doesn't always foster the best development process, and can impede building team momentum or trust.

However, even the most amazing vision can falter. A game designer might pen an exemplary game design document, lauded by all and implemented flawlessly, yet the game could still be lackluster. You still need to divine what is special about your game. What is the thing that makes the game stand out. While rapid iteration with tools like ChatGPT can expedite the development of such intuition and valuable experience, human vision is still guiding the ship.

Take, for instance, my most recent employer, Facebook/Meta. They excel at executing incremental improvements. Their culture and incentive structures are primed for enhancing an established direction. Equipped with data analysis tools and continuous builds, they can carry out large-scale tests to drive their metrics. Snapchat to Stories. TikTok to Reels. Most recently recovery from the Apple Ad tracking change. Despite this strength, they occasionally lack vision, noticeable in the gap between "how do we measure it's being successful now" and "this is what good looks like". A prime example from this week is the "magical" Vision Pro UI developed for the Quest Pro. Now, we have a tangible benchmark of what Apple thinks 'good' is.

Okay, actual Vision Pro thoughts...

I'm glad that the Vision Pro exists and will hopefully launch next year. While its high price limits accessibility, it presents an opportunity to gauge whether these computationally expensive experiences do cross the line of want to need. All the reviews have said the screens really are that good and the interface is “magical”. I still wonder “to do what.”

Zvi Mowshowitz stated eloquently on his Substack:

Such bold attempts excite me. I appreciate the vision to set a high bar and the ambition to reach it. Apple's significant silicon advantage, combined with their ethos of software/hardware integration and focus on product quality, will help determine if this is the genesis of a new computing platform, or merely a "trifle".

Text edited by ChatGPT (I noticed that it completely stripped out any personality in the first draft and had it rewrite a few sentences and that was much better in maintaining my voice).

ChatGPT prompt for Midjourney : "Generate an abstract image depicting a team of game developers standing on the edge of a cliff, looking out onto a horizon where the sun is rising. The sunrise represents their vision and the path ahead, with vibrant rays illuminating a landscape filled with various game-related elements like consoles, controllers, and characters. Show the developers equipped with tools, signifying the technology they're building. Overlay the image with a subtle grid or network pattern to represent the interconnectedness of their work and the digital realm they're influencing."

I had to edit it down to get more of what I was envisioning by trying to keep it simple, since there was so much detail in the generated prompt: "Generate an image depicting a team of developers hiking UP on a hillside . They have NOT crossed over the crest and they do not see what's on the other side. On the other side, there is horizon where the sun is just rising. The landscape is in a mountain range.”

(Image by Midjourney)

Last Monday, my co-founders and I undertook a valuable exercise: answering the "50 Questions to Explore with a Potential Co-Founder". This exercise proved invaluable for understanding our work styles, personalities, and experiences. If you're contemplating a startup, I can't recommend this list enough. One question that really got us talking was "How many hours per week are you willing to work, and for how long? What sounds ideal? What sounds unbearable?". Two of the three co-founders expressed that 'hell' for them was working without making progress, while the third described hell as 'crunch'. This prompted us to discuss what crunch meant for each of us.

The main concern was the idea of constructing a compromised product merely to meet a deadline. This concern resonates deeply with my approach to leading teams. I've previously discussed the importance of resilient (rather than accurate) schedules, the value of iteration, and the significance of clear ownership. To this list of my core development beliefs, I'd like to add the importance of making decisions early when there are more options available. The more time you have, the more paths open up to you, enhancing the possibility of delivering a great product. Crunch usually occurs when you've made promises to publishers, customers, or investors without allotting an adequate timeframe to fulfill those promises. As attributed to renowned game designer Shigeru Miyamoto, "A Delayed Game is Eventually Good, but a Bad Game is Bad Forever". Understanding the cost of decisions and recognizing when to make them is a hallmark of an excellent producer/program manager. These individuals, who can foresee how all the pieces will come together and the confidence level of each element, are invaluable.

Those contemplating a career in the games industry often ask me about the culture of 'crunch'. The most intense crunch period I've experienced was during the final few months of a game project, when everyone on the team was required to be in the office for six hours on both Saturday and Sunday--regardless of whether their presence was actually necessary. It was an attempt at team solidarity! I believe this routine lasted for about a month to get us back on schedule. As you might expect, it was as demoralizing as it sounds. The crunch period was necessary to achieve the vision that the creative director had (or thought they had) for the game.

However, on other projects, I've frequently worked until 2 or 3 AM out of personal choice. On my first project, I used to work until 9PM, go play hockey, and come back and work until the sun rise (oh to be young again). A sense of "let's see what I can build" magic propelled me forward. This internal drive to make the game exceptional is, in essence, the core spirit of game development. There's always an extra touch of magic or polish you can add to a game, but this is only possible once the core game is built. Reflecting on my conversations with the Blizzard team about Warcraft II or reading about Tears of the Kingdom, it's clear that they completed the core game a year before launch and then continued to polish it. Yes, polishing a game truly takes that long. Another factor that makes any extra work feel less like a crunch is giving the team the autonomy to make those changes that really enhance the game. Games have a unique quality regarding the 'feel', not just the functionality. They need to be intuitive and straightforward. This holds true for abstract games that need clarity and immersive simulations that require extra time to smooth out the rough edges where the simulation doesn't align with reality or expectations. Sadly crunch is usually not to do this part but to just get the core game complete. Have you already lost from a game experience stand point when you have to crunch in this way?

The ideal strategy to handle crunch is to avoid it altogether! However, if you do find yourself in such a situation, the first step should be to extend the deadline AND (not OR) conduct a sincere reevaluation of what you can accomplish within the new timeline. Regardless of whether you can adjust the deadline, I believe it's crucial to ruthlessly trim down to ensure the best possible quality of the final product, thereby motivating the team. The goal isn't merely to prevent the product from being poor but to strive to create the best product possible. Remember, the final product is by what you will be judged.

In my opinion, the games industry grapples with overly ambitious creatives or a shortage of individuals who can realistically assess project scope or timelines. Misaligned expectations often lead to unattainable goals, as tasks invariably take longer than anticipated. Seasonal constraints or a specific launch window can impose hard deadlines. Nevertheless, it's crucial to continually evaluate progress and hone down to the essence of what you have, rather than merely ticking off a checklist.

Text editing by ChatGPT (GPT-4)

The image above was created by Midjourney with this image Prompt generated by ChatGPT: "Create a stylized image of a digital stopwatch ticking down on a chaotic desk filled with game development notes, sketches, and coffee mugs, emphasizing the race against time in game production. The background subtly transitions from vibrant colors indicating creative energy at the start, to a darker, more stressful palette indicating the pressures of crunch time. However, include a beacon of soft light hinting at hope and resilience amid the chaos."

Bonus Content

The Midjourney image generation have become a very fulfilling part of my process. And I’m sharing a cool piece of art when I tried to create my own image prompt and got some cool pieces but not as effective of representing the post: “stylized, almost abstract version of an exhausted person slouched over typing on a computer with a monitor of video game development (working in an 3d editor). They have a huge dream bubble - one dreaming of leaders who did a better job laying out a hike through a jungle to a beautiful city.”

(Created by Midjourney)

Let's try something new. In a departure from my usual advice-driven posts, I'm choosing to share a personal glimpse of my current circumstances -- a narrative that I hope will resonate with and perhaps even inspire some of you.

After concluding a nearly five-month hiatus from work, I am now navigating the transitional phase of getting back to work mode. This extended break often felt like a prolonged weekend, filled with tasks such as moving homes and setting them up rather than a traditional vacation. During the first month after stepping away from my work, I grappled with residual burnout and struggled to recalibrate my mindset from work mode to rest mode. Although I've pulled long hours at previous jobs without issue, the question that nagged at me towards the end was, "Is this current effort worth it?"

A creeping sense of apathy set in as I found myself increasingly uncaring about the quality of certain aspects of my job, rooted in the belief that they were inconsequential given the machinery of the system around me. Though there were elements I still cared deeply about (like the people), I could feel my passion ebbing away and knew I needed to step away before I lost love for even those aspects. In the end, I realized my time wasn't spent making things great, but preventing things from sucking and that is draining if there's no end in sight.

Months later, as I reflect on my recovery process, it strikes me that a significant part of it involved shedding the unceasing sense of crisis. Constantly being in a state of alert, braced for the next problem to tackle, had taken a toll on me. Achieving a state of relaxation and proactive thinking proved more challenging than I'd anticipated. I recall attempting to unwind with a new game, only to discover that my mind was so fatigued, it couldn't even grasp the game's mechanics -- not even for fun! It was Covid, a forced period of inactivity, that served as a reality check. During my battle with the virus, the physical exhaustion left me no choice but to give my brain a break. In an unexpected twist, this felt like a final cleanse, ridding me of residual work stress. Post-recovery, I found myself hungry for a new challenge -- a feeling I hadn't experienced in a while.

Over the past weeks, I've immersed myself in an array of AI news, articles, Substack and Medium pieces, and academic papers, gearing up for my forthcoming professional adventure1. Rejuvenated, my mind is once again eager to soak up new ideas. The AI/ML sector is rapidly evolving, buzzing with hype especially among major tech companies trying to outdo each other, not to mention the new kid on the block, OpenAI. I find myself efficiently sifting through the influx of information, delving deeper into the crucial 25% to stay informed. Initially, this figure was around 90%, but I've since learned to distinguish between the meaningful content and the hype or repetitive clickbait.

This past week, I returned to programming after a hiatus of over six and a half years. To shake off the rust, I chose a project unrelated to AI to focus solely on regaining my coding rhythm. Despite overseeing teams handling full-stack web development across multiple roles, I'd never done it personally. I decided to leverage tools like ChatGPT and GitHub Copilot as my programming assistants, partly to assess their capabilities and partly to gauge the scale and type of engineering team I might need to hire, given the buzz around AI amplifying engineers' productivity tenfold.

After a few days experimenting with these AI tools, it's become apparent that I will never willingly code again without a tool like ChatGPT as a reference or an AI assistant to help with boilerplate code. However, I also recognize the current limitations of AI programming tools. I don't envision them rendering human programmers obsolete any time soon. The process still requires a human touch -- judgment, creativity, and personal style -- to define what I want to create and how to piece everything together for a seamless future. I think figuring out WHERE you want to go is critical and not sure LLMs will do it, as they are done now. Perhaps as generative AI coding products improve, they might require minimal human intervention, significantly less than the current 50% I've experienced. This foray into AI-powered programming has sparked some preliminary thoughts about human value in an AI-dominated world.

Despite my transition to full-time work starting next Monday, I hope to maintain my weekly writing schedule. One of my co-founders and I will kick off our collaboration then, though likely spending half the day discussing/being distracted by Apple's latest announcements. Not immediately, but in due course, you can expect to find startup stories gracing this newsletter.

Text edited/cleaned up by ChatGPT but I’ve noticed that it strips away some of my style and dulls down some of my points. I had to redo a few lines to make it feel like me, instead of a robo-Art. Also I’ve heard from a few people (and Twitter) that ChatGPT-4 feels worse recently.

Image generated by Midjourney from this prompt by Chat-GPT. Had to re-run the prompt twice completely: ""Generate an image depicting the journey from burnout to rejuvenation. Show a figure burdened under a heavy cloud in a dull, monochromatic environment, signifying the burnout phase. Gradually transition the image to show the same figure slowly shedding the cloud, the environment transforming into a vibrant, colorful one, symbolizing the recovery phase. The figure should eventually stand unburdened, looking ahead towards a bright horizon, representing rejuvenation and eagerness for new challenges."

(Screenshot of Tears of the Kingdom - crazy mech creation shared on TikTok)

Currently, the gaming world is buzzing about "Tears of the Kingdom," the latest entrant in the Legend of Zelda series from Nintendo. Players are enthralled by its open-ended approach to problem-solving within the game. Perhaps you construct a platform to surmount a barrier, or you opt to fly over it, or you decide to move a barrel of dynamite next to it to blast it apart, or create a mech robot to walk over the barrier (like screenshot above), and so on. Offering players this kind of agency in problem-solving injects a significant element of fun into the game. This is a departure from many games that are laden with specific puzzles intended to be solved in a particular way, such as navigating to a certain location, utilizing an item obtained in a previous room, and confronting the carefully tuned adversaries thus unleashed. On the contrary, games with this open-ended style are gaining popularity (not just the open world ones), like Elden Ring, the consensus game of the year in 2022.

(Screenshot of System Shock)

This concept of gameplay aligns with my roots: "emergent gameplay." One of my career lottery wins was being fortunate to launch my career at Looking Glass (LG) Technologies right after college, working on the cult hit, System Shock. LG has been lauded by Polygon for trailblazing "player choice" and "immersive sim" that inspired the gameplay found in “Tears”. I had the privilege to work with and form close friendships with two pioneers in this space, Doug Church and Warren Spector. They're the kind of designers who redirect the spotlight from themselves onto the team but they really pushed this style of player agency games. Their leadership made sure we focused on player agency as we were building the game.

Looking Glass introduced many environmental interactions found in the new Zelda in their game, Thief: The Dark Project, through a system known as Act/React. Objects were treated as materials that could react to various stimuli. For instance, a wooden chair could catch fire when a lit torch or a flaming arrow is applied to it, which could then set the wall tapestry alight. Sound familiar to "Tears of the Kingdom" players?

While working as the Executive Producer on Deus Ex, Warren Spector aimed for each puzzle or challenge to have three potential solutions -- a hacking solution (tech), a combat solution (fight), or a stealth solution (sneak). This approach encourages designers to contemplate multiple paths or options, fueling the diverse narratives that Warren loved -- stories where two players could exchange "That's not how I did it" or "I didn't even know you could do that."

Not only were my professional formative years at LG shaped by these player-driven games, but they've also had a profound personal influence on me. In 2003, I faced a decision: move to Seattle to work at a more linear-game driven company, or join Warren in Austin, Texas, to develop more open-ended games. I chose the latter, driven by my passion and conviction for this type of gaming. It was a life changing decision as I met my wonderful wife there.

The triumph of "Tears of the Kingdom" provides a lesson for game designers: relinquish control. The focus shouldn't rest solely on the designer's ingenuity in crafting a brain teaser of a challenge or puzzle, but rather on constructing an environment where players can solve problems in their own unique way. Indeed, it's tougher to design an experience that allows for multiple problem-solving routes, since it involves considering a range of possibilities. The goal is to create a challenge that's engaging enough to entertain while being broad enough to spotlight the player's creativity. A notable aspect of "Tears" is that it enables players to scale any climbable wall, regardless of its height, or to swoop down from the sky from any point. This necessitates inventive world-building, like creating mountain lips that can't be climbed over and adding more underground or cave elements. But going to many places feel like you "cheated" by your own ingenuity. Similarly, the game features a power that allows players to ascend through ceilings -- a mechanic that could disrupt the game unless carefully designed around. Thus, while designers still craft puzzles, the array of tools provided to the player changes the environmental cues given to the player.

I remember one Shrine puzzle in "Tears" that I'm pretty sure I didn't solve "correctly." Faced with a body of water and a high wall on the other side, I used a white Chu Chu jelly (ice equipment) to create an ice platform and bypass the puzzle -- despite there being no ice equipment in the nearby area (which are clues to the 'safe/straight forward' solution). This begs the question: did I cheat? What's the purpose of a puzzle? Is it to figure out the designer's intent? Or is it to navigate a tricky situation with provided tools, with the puzzle's resolution being the real challenge, regardless of how it's achieved? Was it fun? Yes. Did I feel clever? Yes.

Finally, I fervently hope that these "immersive sims" continue to progress and expand. Are more game designers ready to bravely invest more time in understanding how to craft experiences offering multiple options? It may seem like more work through iteration and much more extensive play testing, and players may only see part of the effort and cost involved in the game's creation. And given the sky-high budgets of modern games, it's disheartening to think that game teams might lack the necessary trust to experiment and learn. However, if I need to convince businesspeople that this path is viable, I'll make a different case -- we live in the era of TikTok, YouTube, and influencers. People adore boasting about their unique, innovative exploits of how they "beat" or "cheat" the game and then SHARING IT. While a straight and linear game may offer limited replay value, a collection of tricks and cool moments that say "look at what I did" can provide fantastic viral marketing opportunities. Ultimately, it's the players who win, as they benefit from the awesome games that we can create. It truly becomes THEIR game.