Optimizing for Time: Dark Matter (Filed under DevEx)

Optimizing for Time: Dark Matter (Rubric #DevEx)

I watched with great interest. report frankly DPE Summit Karima Nakada (Karim Nakad) from the banned company Meta in Russia. I liked the name, which refers to cosmological theories that explain the state of our universe through dark matter. Here, this metaphor refers to the fact that we often focus on the visible spectrum of development: improving IDEs and speeding up builds, but ignore the huge time spent on daily routine. In fact, the speaker calls “dark matter” the invisible load on employees: reading corporate chats, stand-ups, writing documentation and helping colleagues. Measuring these latent activities helps to identify the true reasons for the slowdown in technical teams.

I especially like the Meta performance metrics that I’ve talked about in other posts, such as the Meta. parsing “Measuring the Impact of AI on Developer Productivity at Meta” But this talk focuses on metrics. Separately, Meta uses aggregated indicators for process analysis, strictly prohibiting their use for individual evaluation of engineers in order to avoid cheating. The metrics themselves are.

- By Tool -» TSD (Time Spent Coding per Diff): reflects the coding time per Pull Request, allowing infrastructure teams to evaluate the effectiveness of developer tools. **- By Intent -» PCT (Percent Time Coding)**Shows the proportion of pure programming in relation to all other tasks, helping to reduce the number of unnecessary meetings. Metrics are important for product teams - By Workflow -» WTS (Workflow Time Spent): tracks the time spent on a particular process (for example, incident management) To find bottlenecks when switching between systems.

• By Value -" but here they did not show metrics, but gave a metaphor about the map and about moving in the right direction. (No matter how this direction is measured)

As far as I'm concerned, I like the fragmentation slides of the day. The slides show how an engineer’s day breaks down into dozens of short switches between IDE, chat, calendar, review tools, notebook, and SQL. Interestingly, Meta offers to look not only at raw coding time, but also at coding intent time – that is, consider engineering work not only minutes in the editor, but also related trips to chat, wiki, SQL and documentation. Otherwise, it is very easy to optimize the wrong place.

I found it useful to have ideas for engineers to protect long focus units, group communications, jam non-critical notifications, and stop thinking that “real work” only happens inside an IDE. And it is useful for managers to calculate the cost of context switching and compare it with the cost of a slow build; separately look at the team focal time, the structure of the calendar, the volume of asynchronous noise and only then twist the tools. Otherwise, you can accelerate the pipeline and still lose the bandwidth of the team.

#Engineering #Software #Bigtech #Productivity #Management #Leadership #Processes #AI #ML #Architecture #DevEx #DevOps