- [2022\_10\_14\_agile\_in\_1000](https://redpenguin101.github.io/html/posts/2022_10_14_agile_in_1000.html) > Hi [@neumann](https://clojurians.slack.com/team/U5FV4MJHG)[@nate](https://clojurians.slack.com/team/U0510902N), I just wanted to share that I've been thinking a lot these past days about the tracer-bullet approach that you kept mentioning in the podcast. I'm looking into [Datastar](https://data-star.dev/) these days and for that I needed a server-sent-event endpoint. I've never built one, so I thought to start with a super embarrasing and leaky  tracer-bullet and it helped me significantly to move forward. I thought to use writing as well next to that and the combo of writing + tracer-bullet worked really well. I think without the tracer-bullet approach, I'd be stuck in writing, I'd never be able to write anything close these posts. But with the tracer-bullet, stopping with an embarring solution felt ... productive. So thanks a lot  >  keep thinking about Alberto Brandolini & DDD saying that: "software engineering is a learning process, working code is a side-effect", with trace-bullets there to maximize learnings, especially in those fuzzy beginnings > “We have to make decisions on how to structure entities, how to define the correct boundaries, and how to define the table schema so queries can be done efficiently. There are many questions to answer, and unfortunately, we have to make a lot of critical decisions when we know the least about the system we are building.” > > “And we will get some of them wrong. That’s inevitable. If you live under the illusion that you just need to spend enough time crafting your perfect data model to prevent this, you are wrong. Sorry.” > > – Martin Dilger, Understanding Eventsourcing