If you only read one thingThe hardest part of building software is not coding, itâs requirements (24 minute read) âThe hardest part of building software is not coding, itâs requirementsâ Especially because no-one wants to do requirement engineering. Product managers want to play with Figma, developers want to implement user stories. All play no hard thinking. And a finger-pointing blame game. âAI canât create software, only codeâ Of course the article is wrong; it thinks AI canât do requirement engineering. Au contraire! Etienne made a great point on our CTO Show. AI writes better tests and error handling because while you know 10% of the code, it knows all the code. It knows much more edge cases, much more about people and will do much better requirement engineering.
Stories Iâve enjoyed this weekSoftware engineers hate code. (10 minute read) Interesting thought. âThis is the best-kept secret of the software engineering profession: engineers hate code. Especially code written by other people.â This drives microservices, this drives the always new frameworks. And people leaving you for a new company where they can do Greenfield development. âRolling green fields forever!â Do Javascript developers hate other peoples code more than Java developers (in JS there are much more rewrites and new frameworks). âOnly one thing can overcome engineersâ hatred of code: their love of writing code.â https://www.dancowell.com/software-engineers-hate-code/ ChatGPT Drops About 10% in Traffic as the Novelty Wears Off (10 minute read) So itâs not looking like ChatGPT will take away all of the Googles traffic - yet. 50% of the things I search on Google I now ask ChatGPT often with better results. âWhat time is 8am PST in Berlin time?â â I get an answer and not something I need to parse for myself. Perhaps the novelty has worn off, perhaps itâs the Gartner hype cycle; perhaps people use more local LLMs perhaps Bard and Bing had an impact. https://www.similarweb.com/blog/insights/ai-news/chatgpt-traffic-drops/ Why Shopify Elevated the Non-Manager Career Path and Ditched Meetings (9 minute read) âI think the most important thing we did [..] was to change the default answer for a meeting invite from being a âyesâ to a âno. [..] This had a massive boost on productivity. The average time people spent in meetings went down by a third.â Meetings are great, and can speed things up if done the right way, but sadly 99% of people use meetings wrong. So itâs better to say no. (Send me an email if you want to know how to do them right). https://creatoreconomy.so/p/kaz-coo-shopify-craft-and-no-meetings Are bugs and slow delivery ok? (8 minute read) I disagree. For years now, I have been a strong proponent of a zero-bug policy. For two reasons: If you accept bugs, you set a low standard and culture for development teams. The belief that quality and craftsmanship doesnât matter sips into everything. Second, bug fixing takes a lot of time. Testing the bug, creating a ticket, scheduling the bug against your roadmap, understanding the code, writing a test, fixing the code, retesting the fix, reading the dozens of comments on the ticket, closing the ticket. High quality and craftsmanship reduces a cognitive load and chaos and streamlines development. See also âExcellence is a habit, but so is failureâ https://awesomekling.github.io/Excellence-is-a-habit-but-so-is-failure/ https://journal.optivem.com/p/are-bugs-and-slow-delivery-ok Feature Flags : Theory vs. Realit (13 minute read) Iâm a fan of feature flags. There are many good reasons to have them; the main reason for me is to decouple releases from feature availability. You can ship code and have the features available to your customers at another time. This makes releases easier and allows for grouping of relevant features independently of when they have been released. This is core to be able to do feature marketing and reduce the cognitive load of your users. Send them an email with the new features, create a feature tour on your website. With feature flags, this is easy. With releases, it is difficult. And product and feature marketing are core to your success, something most startups I see forget. https://bpapillon.com/post/feature-flags-theory-vs-reality/ Create an advanced search engine with PostgreSQL (23 minute read) Iâve written a Blog Post called âJust use Postgres for everything.â Putting Postgres in the center of your architecture makes operations much easier. Itâs not the best tool in each class, but reducing the number of tools from 5 to 1 makes a difference. Most can be done with Postgres fulltext search instead of Elastic. https://xata.io/blog/postgres-full-text-search-engine How to Handle Conflict on Remote Teams (15 minute read) Companies are bad at handling conflict. One group of companies do not like conflict, there are only happy shiny people. No real progress is achieved because necessary conflict is avoided. The other group of companies love conflict. No real progress is achieved because there is infighting everywhere. Remote makes this even more difficult. Very good tips on handling remote conflict. https://www.helpscout.com/blog/remote-team-conflict/ mshumer/gpt-prompt-engineer (9 minute read) Meta prompt engineering. âSimply input a description of your task and some test cases, and the system will generate, test, and rank a multitude of prompts to find the ones that perform the best.â https://github.com/mshumer/gpt-prompt-engineer Harvard ethics professor allegedly fabricated multiple behavioral science studies (0 minute read) This has also been replicated in studies, so itâs not just an anecdote. The same goes in software engineering and blog posts. Large companies write about doing X, then doing Y. And all the small startups follow idiotic ideas. I canât count the number of CTOs Iâve met who failed with the Spotify model. Branchless Programming: Why âIfâ is Sloowww⊠and what we can do about it! (20 minute read) As a coder of 40 years, Iâve learned a lot in this video. A totally new way of thinking for me. https://www.youtube.com/watch?v=bVJ-mWWL7cE NETSCAPE AND SUN ANNOUNCE JAVASCRIPT (1995) (5 minute read) Oh my! 1995. https://web.archive.org/web/20070916144913/https://wp.netscape.com/newsref/pr/newsrelease67.html AI Tests Into Top 1% for Original Creative Thinking (6 minute read) â New research from the University of Montana and its partners suggests artificial intelligence can match the top 1% of human thinkers on a standard test for creativity." Iâm out. https://www.umt.edu/news/2023/07/070523test.php New study finds an unstructured 5-minute break can help restore attention (15 minute read) âThey found a 5-minute break from thinking is all you need to get your concentration back. There is no need for a walk along a river, or a lengthy video of bamboo forests swaying in the wind (although that could be nice). A five-minute total break will do the trick.â But not reading my newsletter! https://medicalxpress.com/news/2023-07-unstructured-minute-attention.html Steve Ballmer Laughs At The iPhone (1 minute read) *âIt doesnât have a keyboard, it doesnât make a very good email machineâ. And it was too expensive with $500. Your assumptions will kill you and your company. Join the CTO newsletter! | |
