Effective Criticism

I’ve received a lot of criticism during my career and have given a lot.  No product, process, or team improves without effective criticism.  And there are a few things I’ve learned about receiving criticism and about giving criticism.

On the receiving end, it is important to realize that you are not your work.   The moment your work leaves your body, it becomes a separate entity, detached from your identity.  And you can look at it, others can look at it, and you can all discuss and criticize it, without it affecting you.   Let go of your ego and allow yourself and others to evaluate your work critically; this is the best way to learn and grow your skills.

On the giving end, I always try to criticize the work, not the person.   Bill Gates was great at this; he could be forthright and very critical, but it was always about the work — “This is the stupidest f&*king thing I have ever seen”.   It’s painful, but it is not about you; it is about the work, and he was often right!

And when giving criticism, I have learned three lessons from a set of colleagues (Jonathan Roberts and Steve Hooper from Ignition days come to mind; they always offered great criticism):

• Always offer to help.   If you are going to criticize, you need to be willing to pitch in and help.  You will find that your criticism is much better accepted if it comes with an offer of help.
• Always offer proposed solutions.  Don’t just point out flaws; that is annoying. We all know our work has flaws.   Bring solutions to the table.
• And the gold standard of criticism — implement a better alternative.  Some people may not take this well, but if you do it in the right spirit and give it freely without strings attached, then the recipient should be able to set aside their ego.    

DOGE

Last week, I was critical of the DOGE effort, yet again.  

I’ve suggested alternative goals for DOGE in the past — we should focus our best industrial talent on driving economic growth in key industries.   Lots of people have had ideas for economic development:

• Here is a comprehensive playbook on the topic — I don’t agree with all of these, and it doesn’t go far enough on skilled immigration and education, but at least people are discussing ideas.  
• Alternatively, you can draw inspiration from various thinkers who advocate for an “abundance agenda,” such as Ezra Klein.   

There is a wealth of thinking on how to refocus on economic growth — let’s get moving!  If I were Commerce Secretary, rather than all this tariff nonsense, I’d have a playbook pulled from these sources above and I’d work my way down it. 

And apparently, Elon has finally come to this conclusion as well: 

I have come to the perhaps obvious conclusion that accelerating GDP growth is essential.

Anyway — one particular idea for growth that I would personally love to work on — we need many more people working in the physical sciences — materials, geology, mechanical engineering, electrical engineering, chemical engineering, etc.  Hadi Partovi and the Code.org team have done an outstanding job promoting computer science literacy.  We need the equivalent for physical sciences — for atoms, as Code.org has done for bits.   I’d love to work with anyone on this.

Microsoft AI

Microsoft has spent a large amount on AI and has gotten much great press.  But they are falling behind on usage.  I chatted with some friends about Copilot’s lack of utility — Copilot is tepid and lacking in utility, even in domains where Microsoft ought to be a clear leader, such as automating Windows tasks and Office tasks.  I don’t have any insight into what is happening in the company.   But I can discuss an analogous time, in 1995, when Microsoft was scrambling to react to the internet tidal wave.   Microsoft initially had 37 responses internally; everybody and their dog were working on an internet strategy.  Three actions were taken to allow Microsoft to become an internet leader:

• CEO leadership.   Bill delivered a strong message that the Internet was the top priority and that the company would pivot towards it.  
• Focus.  After some fits and starts, the company focused on a single browser strategy with IE, a single server strategy with IIS, and a single authoring tool in FrontPage.  There were also some smaller efforts, but the company remained fairly focused.
• Great product teams.  The company refocused some of its best teams on these internet product efforts.  A large part of the Windows 95 team went to work on IE; the NT team took on IIS; the Office team took ownership of authoring with FrontPage.  The refocusing of these mainstream teams in the company was not without cost; other commitments and plans had to be set aside, and this made people angry.  But it was the right move.

Certainly, Satya is providing CEO leadership on AI.  I wonder about the other points — in particular, what great product teams inside the company have dropped their current work and refocused on AI tools and apps?

US science leadership

I am completely going to overindex on a couple points in this study — Velvet worm slime transforms from liquid to fiber, and back again, a discovery that could inspire recyclable bioplastics — from McGill University.

1. It is fascinating to see how AI tools have impacted the analysis of biomaterials — Alphafold was one of the key tools used.  
2. The research was done in Canada and Singapore.  Not the US.  A harbinger of what is to come as the US tears down its research funding and research institutions.  

We need to invest more, not less, in science leadership.  As mentioned above, I’d love to work with anyone on building grass roots support for physical science education.

Software is eating the world, part 26

This is not a criticism; just an observation.

Notboring argues that technology is going to chew through almost every single industry, rewriting the rules of each as it goes.  Moore’s Law is a powerful force — it is going to come for every industry, one way or another.  

And Notboring argues that venture-backed startups are a critical part.  If there is a chink in China’s industrial armor, it may be the lack of a thriving startup market.

A short note on development tools

One man’s guide to vibe coding — Sam offers a very structured approach to working with generative AI tools to help with coding — a methodical stepwise approach to building requirements, project structure, an implementation plan, and individual elements with testing.  This resonates; my earlier attempts at letting the tools write the app with only feature goals as a direction have all failed.   I need to adopt this more disciplined approach.

The greatest thing about AI coding tools is that I can skip the frontend JavaScript morass.   I can use the AI tool as the complete answer and just target html:

AI's out here, a gift from the heavens (or at least from Sam Altman's nerd fortress) ready to write your shitty little to-do app in five seconds flat. It can churn out pixel-perfect HTML, debug your fuck-ups, and probably even wipe your ass if you ask nicely. But no, you're still humping your frameworks like they're the last lifeboat on the Titanic. What the hell is wrong with you? Are you that addicted to 10,000 dependencies and a build process that takes longer than your last failed relationship?

Sam Altman's AI army is laughing its silicon balls off while you're knee-deep in React's virtual DOMshit, praying your app doesn't choke on its own bloated corpse. This isn't progress. It's a fucking tragedy. You've got a shiny new Ferrari in your garage, and you're still riding a rusty tricycle with a flat tire. Grow the fuck up.

Closing observation

I often discuss why I write — how critical daily writing is to clear thinking.   And that is all true, but there is some deeper truth in this from Joan Westenberg:

Every creative person has two wolves inside them. One wants to start a newsletter. The other wants to fake their own death.