Overthinking (2021)

Amazon has an internal tool called &quot;Forte&quot;, a tool used once a year for employees to provide &quot;anonymous&quot; feedback for one another. One piece of feedback that cropped up multiple times for me, from multiple people, was that I could improve in &quot;bias for action&quot;, akin to analysis paralysis mentioned in the article.<p>At first, I got a bit defensive ... and in response, I ended up running an experiment, delivering code &amp; written documents that — inside my head — felt incomplete, unpolished, not quite at the &quot;bar&quot;.<p>The feedback following?<p>Overwhelmingly positive.<p>I had anticipated that my peers and leadership would notice a drop in quality. Instead, I was commended for speed of delivery.

This made me think of science and academia. One of the things that seems to distinguish productive people in science is the balance between “doer” mentality and “thinker” mentality.<p>Too much “thinker” mentality and the project never goes anywhere. Too much “doer” mentally and the project moves but may go down an unproductive path.<p>Perhaps an analogy is that “thinker” and “doer” mentalities work together like a stochastic gradient descent algorithm.<p>The “thinker” mode tries to calculate accurate gradients, but never moves towards the goal.<p>The “doer” mode takes a step towards the next iteration, regardless of whether you have an accurate gradient already.<p>Balancing the two correctly can give beautiful momentum dynamics that steers towards your goal.

A refreshingly good take on paralysis by analysis. I have suffered from this in the past as well. I found that having projects that are otherwise meaningless that I can hack together helps. It’s almost like a mini hackathon where the only thing that matters is the end product. Maybe I’ll choose something completely out of my professional life (eg woodworking with scrap wood) or write something in a language nobody around me likes because it’s “ugly.”

&gt; I&#x27;ve also noticed that, up to a certain point, the smarter a person is, the more it has to be apparent in their work. Every algorithm needs to be perfect, every function needs to be side-effects free, every data structure needs to be the fastest, and every best practice needs to be followed.<p>Many engineers are somewhere on the Asperger&#x27;s spectrum, as Temple Grandin tells her Googler audience in [1]. Overthinking is a prime symptom of it. I&#x27;m disappointed to see that not even mentioned in this article.<p>There are some engineering practices that, unfortunately, amplify this rather than tamping it down. Code reviews, in particular, can do that; a reviewer gets points by nitpicking (&quot;you could have done that in one line instead of two!&quot;).<p>[1] <a href="https:&#x2F;&#x2F;www.youtube.com&#x2F;watch?v=IA4tE3_2qmI" rel="nofollow">https:&#x2F;&#x2F;www.youtube.com&#x2F;watch?v=IA4tE3_2qmI</a>

<i>We all know that &quot;The perfect is the enemy of the good.&quot;, still we lose countless hours trying to make the perfect design, the perfect</i><p>Yesterday I helped a guy to register his little perfume company’s trademark on a government platform. We filled in all the fields and then unexpectedly (to me) it just asked for a picture. I thought okay, maybe we’ll get back to it after a designer makes a logo.<p>The guy pulls his iphone and creates a note. “Good fonts”. Three enters, &lt;companyname&gt;, enter, parfum. He asks how to make first line bigger, I show him. He indents with spaces. I point out that alignment is not exactly centered. “Looks okay”, he takes a screenshot, crops it to a rectangle and sends it to me to upload into the form, and to his chinese partner, straight into production.

Thing that helped me with avoiding this kind of problem was learning about wabi sabi[0]. A mindset of accepting and finding beauty in imperfection.<p>[0] <a href="https:&#x2F;&#x2F;en.wikipedia.org&#x2F;wiki&#x2F;Wabi-sabi" rel="nofollow">https:&#x2F;&#x2F;en.wikipedia.org&#x2F;wiki&#x2F;Wabi-sabi</a>

I penned this short verse a long time ago to help remind me about &quot;Analysis Paralysis&quot; and hopefully try and minimize its effects on me :-) Being reminded of it from time to time, has helped me avoid &#x27;stalling out&#x27; on too many details and better learning to live with imperfect information to make progress in a reasonable timeframe.<p>---<p>&quot;A person who faces many choices, really has one<p>Out of indecision and confusion, they choose to have none&quot;

I like the article. The architecture bit I&#x27;m not so sure about though. I wish in my company the architects actually did <i>more</i> thinking and especially talking &#x2F; negotiating with all the other architects of adjacent components. That would really help. Instead they work hard to fix minor problems and build walls in between the components... or yet another middleware-generator-middleware-wrapper.

There is an assumption here is doing anything is better than doing nothing. Inaction is not inherently inferior to action. Sometimes, if not most of the times, doing nothing is exactly what you need to do in response to something. If your brain prevents you from doing something, maybe it&#x27;s because it has a point? Trying to find a shortcut by generalizing will do more harm than good.

A big part of learning this is <i>not</i> overanalyzing past failures. Check and see if you missed something major, but then don&#x27;t dwell much on the details; perhaps anything you would have done would have been doomed to failure; it wasn&#x27;t the time, etc.

Relatedly, this is also a big reason why I personally struggle to work at large tech companies. The velocity just isn&#x27;t there because you <i>have</i> to be more thoughtful before you act - the environment is less conducive to quick action and less forgiving of mistakes. And you specifically have to be more thoughtful about considerations that are often immaterial to core business value. I realized after a while that certain large company experiences can train you such that you no longer know how to hustle, and that seemed like a considerable invisible cost of the arrangement.

This is a very minor nitpick, but saying you need a specific technology is part of the problem, even if it&#x27;s a really good and appropriately-complicated technology like postgres.<p>90% of the time, you need whatever technology your company already uses, or your team already knows, that solves the problem.<p>If your company already uses .NET and SQL Server, you probably don&#x27;t need node.js and postgres. If your company already uses PHP and sqlite, you still might not need postgres, unless there&#x27;s an identifiable reason to switch.

&gt; Is it better to tweet at 18:00 or 18:05?<p>&gt; Rust or Go?<p>Well that&#x27;s quite a range.

To me, i can translate this article like:<p>- OK stop all the complexity of devops processes, just ssh to the server, then run `docker compose build &amp;&amp; docker compose up -d`.<p>Now i can have my app running in production to serve its purpose. Problem solved.

To measure the meaning can make you delay.<p>It&#x27;s time you stop thinkin&#x27; and waisting the day.<p>—from The Hobbit (1977) [0]<p>[0] <a href="https:&#x2F;&#x2F;www.youtube.com&#x2F;watch?v=2dQ5c5SIYnc" rel="nofollow">https:&#x2F;&#x2F;www.youtube.com&#x2F;watch?v=2dQ5c5SIYnc</a>

&gt; Thus we learn how to think, and we end up with whole countries filled with people who think, think, think but never start doing anything.<p>That could of have used a lot more thought: What is the basis for it? Are less educated people more productive? Is the US ever portrayed that way?<p>IME, there are individuals who overthink - that is, they think for emotional needs not related to the problem. But I don&#x27;t think I&#x27;ve ever encountered an organization where generally too much thinking was going on!<p>I think it&#x27;s trendy to denigrate or ignore critical thought, education, reason, facts ... Our decisions come from somewhere; either we think for ourselves or we are victims of others who will think for you and influence the public. It seems so many trends these days benefit a powerful few and train the public to follow them obediently.

I tend to ship fairly quickly, at a high Quality (but not perfect) level, then go back and clean up, after it’s out there.<p>It’s important that the first release be extremely high Quality. I’m not a fan of lash-up MVPs.<p>I work <i>very</i> quickly, and do good work.<p>I just did that with the 2.0 version of one of my apps. I released 2.0 a couple weeks ago, and it’s at 2.1.4 (I think). I test, and solicit feedback. One of the releases covered feedback on the App Store.<p>It stops, after a few tweaks.

Living in China it&#x27;s quite amazing how the business culture differs from many western markets. People seem to throw themselves in to ventures without business plans, market research or specific costings. I suppose that when the cost of failure is reduced, dynamism results, because reaction times to opportunities are reduced and people are able to take the risk of following a new path. These days, when I think of analysis paralysis, I think of conservative traditional western business mindsets. The worst of which, frankly, seem to be continental European and governmental bureaucracies.<p>FWIW in the last 18 months I recall pitching one major European industrial group requesting specifically disruptive technology for established industries. Considered at the board level, their feedback was unanimously positive: but they could not take the opportunity because it was &quot;too far from existing business lines&quot;. If you thought corporate VC was bad, try that in an old-Europe context...