• 0 Posts
  • 24 Comments
Joined 26 days ago
cake
Cake day: March 31st, 2025

help-circle















  • Well, often they know it´s hard to estimate, but the entire corporate system is built around having things done by a certain date, your time costs money and payments are usually linked to those dates. They don´t really have a choice but to make a planning based on the estimates you give and monitor the progress so they can give the proper level of panic to their bosses. Of course, software has always been a disaster with estimates and attempts to tame the chaos haven´t been that successful.

    I usually make a ridiculously detailed list of all tasks. ¨Add button A on screen. Discuss details: 2 hours. Interface work: 0.5 hour. Code work: 2 hours. Database work: 2 hours. Testing: 2 hours. FAT: 2 hours. Changes after FAT: 1 hour. SAT: 2 hour. Test script: 1 hour. Update documentation: 2 hours. Add button B … ¨ Put it all in an excel sheet and summarize. Most PMs don´t even want to start arguing a list like that, and it seems to make a reasonably good estimate for me.


  • It’s an attempt to get a handle on things and trying to avoid situations such as:

    “Oh, I was struck on that point for the last 3 months. I reinvented the wheel 2 times and now it works.”

    “And now we’re 3 months behind schedule. Why didn’t you ask anybody?”

    “Yeah, I didn’t want to bother anyone. But I did put in on the timesheets.”

    “It says ‘working on project’.”

    And that’s how regular project update meetings get scheduled, and a bunch of messages asking for updates.



  • how you never have anyone who can actually “manage” the AI workers.

    You just use other AI to manage those worker AI. Experiments do show that having different instances of AI/LLM, each with an assigned role like manager, designer, coding or quality checks, perform pretty good working together. But that was with small stuff. I haven’t seen anyone wiling to test with complex products.