Maaret Pyhäjärvi

There’s a teamwork simulation with two conflicting strategies. One centers AI-generating specs from ‘mother of all specs for this domain’ and another centers socializing specs as common foundation. These two are hard to design into one since humans are bad at taking on a wall of text. Bringing down batch size is not this team’s strong skill.

The difference between the mostly-manual pipeline based on process and instructions and the modern kind where the pipeline is mostly automated is in what happens when the schedule is at risk. The first breaks quality. The latter breaks scope.

I've been observing how 2 weeks cut from schedule with the first kind is having an impact, making the difference painfully clear.

How to fail at your work while praised.

Deliver excellent analysis and proposals to a client that has no budget, thus never being able to pay for your work.

Raise the bar of what testers should be capable of, and be used as tool of personal promotion for the potential people to recruit.

10 people interviewed. 1 fit this modern testing profile. Apparently that means they were qualified to grow into the next position in their current company.

Good it is weekend. Could use a break.

Consultant life: received praise on work I had done for a client. Not just a usual good job, but backchannel and front saying the work was excellent. It’s just that the work is proposal so everything I do is at loss. I really dislike this dynamic. Excellent work for free is wasted.

Experienced some of the most embarrassing ‘team’ dynamic today and I think it’s for the wrong use of AI agents. When your entire collaboration collapses to prompt design, it’s five people watching a random output being generated. I don’t often give up but now I’m seriously thinking of it.

I was interrupted for scolding on a women’s day for saying to a man ‘Thank you for repeating the proposal I just made’. For a moment I’m particularly unhappy with it.

Would I say that again - most likely yes. Even when it’s a customer teamwork simulation.

I get that he missed the proposal because his attention was on prompting an AI not on his fellow humans. So not looking for this AI future.

I will be in Portland, Oregon on October 12-14 (and around). Note to self, set up something with the brilliant @DianaOfPortland@agilealliance.social.

With two speakers pre-scheduled, it it acceptable for the first to even ask the second if they can go over time? Thinking of questions where you surprise the other in a moment making saying ‘no’ a difficult response.

Then again, my answer for the day and the audience was no with commentary on how optics of man going over time in women’s day event would work. The talk hit timebox to a minute after this.

Got a new title today in talk intro ‘madam of testing’. I take it women’s day celebration.

Whoever comes are the right people x100. I believe it but I still need it repeated when 4/10 show up to a teaching session with hundreds of people in the queue.

I feel like saying "this is why we can't have nice things", but the condensed quality of this crowd really made it feel like the right people.

While in management, I could afford to not trust. Now in leadership, I can no longer afford that.

There is a requirement in public procurement: “I have tested system-wide solutions that include microservices for an account-based ticketing system for public transport over the past two years.”

If I am a professional tester and I am using (and complaining) about their systems every day for last two years, does that count for testing. Specifically, nothing says I was approved to test it in their test environments, or that I was paid for the work I did, or that the work I did was complete.

For Robot Framework structures, we packaged our linters and agent files as the part we reuse across various test automation repos. Making disabling a central repo thing rather than distributed is what this was designed for.

Hijacked an online session by silently taking visual notes while on screen share. I modeled what is asked, what is said, where we disagree, where we make agreements. I miss times when you could do this on a physical whiteboard but we don’t have any.

Strategically showing emotions. It’s funny how people think I have no choice in expressing frustration out loud with emotion visible just because they choose to be almost too contained. People may move and wake up, and for many things the one thing we can’t afford is complacency.

Showing what I vibe coded last week, this comment from a friend I appreciate: "You could sell this stuff" on one of my four things.

[Blog] 4x improvement, 13x improvement
https://visible-quality.blogspot.com/2026/03/4x-improvement-13x-improvement.html

Delivered a feedback: "You are not producing enough value for the price the client is paying".

This leads to the salary you are being paid.

Which leads to "What should I produce to be worth this salary?"

The response on this level tends to be that you aren't accepting assignments with clear targets, you are negotiating value. And most likely now you did not.

The active, intentional time used on hands-on testing is the only relevant way to talk about how coverage of testing is.

Under label of "testing", we do things that don't grow coverage:

- report bugs, investigate them, and retest after fixes
- set up data and environments, intake information so that we have the knowledge and ability to test
- participate in team activities, including specifications - even leading them.

Due to budget cuts, a tester was moved from 100% to 50% allocation. I kept asking if we cut the work and the value, or only the work. So we analyzed metrics.

We cut the work, and doubled the value.

We still would process 40 changes a month and test them. We would plan based on risks better. And since the work includes less learning of the system on later years, learning got cut from the hours almost completely.

Finding problems outside *this change* was down. But contributing specs was up.

Had one of those ‘managers should explain more how they think and the why’ and found myself responding with pointing out sound of trust they have on us which we rarely reciprocate.