OKRs == delivering the roadmap?

?? Killed by LLM

by Stephan Schmidt

Happy ?? Sunday,

Welcome to my opinionated newsletter,?today I have something different for you.?Recently I’ve started writing booklets for CTOs on topics that I deem very important, but I do not see covered somewhere else. The first is?“Engineering Role Descriptions”?- on why and how to write role descriptions and use them for the best effect.

Best: They are free.?Go get yours!

https://leanpub.com/engineeringroledescriptions/

PLUS This week’s insights

• ? Imprecise Asks Are a Big Waste of Time
• ?? OKRs are just delivering the roadmap ?
• ?? Killed by LLM

Good reading, have a nice Sunday ?? and a great week,

Stephan CTO-Coach and CTO-veteran

Need support as an engineering manager? Thought about?coaching??Let's talk—I helped many CTOs and engineering leaders with growth and making the right decisions under pressure, I can help you too.

??

If you only read one thing

Your Small Imprecise Ask Is a Big Waste of Their Time (4 minute read)

Once a CEO I reported to told me,?“Stephan be careful with what you say in a meeting, someone might take it as a task and run with it, spending a week on something you were just a little curious about.”?The article is about that and beyond.?“A manager asks a direct report “can you do X?” The report spends 10 times the amount of time that was expected to do the task, leaving the manager befuddled”?The article has some good insights, but the article has also some good tips like?“Clarify the expected time investment”.?Must read.

https://staysaasy.com/startups/2023/11/10/imprecise-asks.html

??

Stories I’ve enjoyed this week

Hitting OKRs vs Doing Your Job (17 minute read)

Isn’t this the core of the problem? OKRs in software development mean you are doing your job??“In Engineering, quarterly OKRs can feel like a duplication of product planning. Basically, they say “Ship the Roadmap.””?Something not enough CEOs understand, and torture CTOs - like my clients - to come up with OKRs. Ship the roadmap! A surprisingly deep article on the topic,?“Why are OKRs easier in Marketing? Marketing is closer to project work, while Engineering is product work.”?IMHO CTO should come up with OKRs that help them achieve?their?goals - not someone else’s goals. You don’t have your own goals? Probably because you don’t have a tech strategy. And you don’t have a tech strategy because you don’t have a tech vision. Start there, OKRs are last in line.

https://jessitron.com/2025/01/05/hitting-okrs-vs-doing-your-job/

Killed by LLM (8 minute read)

Remarkable! A long list of benchmarks of LLMs, on coding, math etc. invented by people to check if LLMs can do it, that are stopped because the LLMs became too good. The first victims of the new age.

https://r0bk.github.io/killedbyllm/

Bonkers Comptime (46 minute read)

Comptime?is one of the concepts that you need to learn about, even if you don’t use it or ever use Zig - where it is most utilized. Comptime is different from macros and other similar concepts, as it is the same language, just executed at compile time. You want to calculate one thing once, do it at comptime. But there is more you can do. This example iterates over the fields of a struct - hey Java can do that! - but it’s not reflection at runtime, it’s done at compile time without the need for Reflection. Genius ??

inline for (comptime std.meta.fieldNames(MyStruct)) |field_name| {
   sum += @field(my_struct, field_name);
}        

My only fear would be that it does, like Rust macros, increase compile times a lot.

https://www.scottredig.com/blog/bonkers_comptime/

Can LLMs write better code if you keep asking them to “write better code”? (50 minute read)

Well, can they? Kind of. And also not. Fascinating read and deep dive. The article takes a problem and repeatedly asks the LLM to write better code. The LLM sometimes does, regresses other times and makes random changes changing technologies - and it also hallucinates. It does not get better with a system prompt. You really need to know what you are doing, and for knowing you probably need to be a seasoned senior developer. But with (excellent) good prompting, there is some progress with?“write better code.”

https://minimaxir.com/2025/01/write-better-code/

The Evolution of SRE at Google (23 minute read)

A very deep trip into how Google SREs work. If you have an operational challenge, a good read. Do you know what STAMP is? I didn’t. " We have adopted the STAMP (System-Theoretic Accident Model and Processes) framework"?- the article goes into details. Worth a read, too many CTOs neglect site operations, or do not know about it because they only have a developer background.?Small side story: Working at ImmoScout24, a large website in Germany, the biggest mistake as a development manager I made, was not sitting a month with operations. It would have made my next CTO job so much easier. What mistake are you currently making?

https://www.usenix.org/publications/loginonline/evolution-sre-google

Conscious unbossing - 52% of Gen-Z professionals don’t want to be middle managers (10 minute read)

You struggle with getting new middle managers. It takes time to get senior developers into management roles, when you need that middle management (trick: Every senior manages an intern to see if it is for them). Now this study says 72% of Gen-Z professionals don’t want to manage other people, with?“69% of Gen-Z say middle management is too high stress, low reward”. As the IC role will change with AI, so will the management role. Hope for Gen-Z that means they get what they want - though my skeptical developer mind doesn’t think so. For you: Even harder to get new middle managers.

https://www.robertwalters.co.uk/insights/news/blog/conscious-unbossing.html

Want the newsletter one day early??? Free to your Inbox every Sunday???https://ctone.ws