Another important observation: in the experience of classic systems, code is a useful means of architectural description for only the simplest of systems: the code is the truth but not the whole truth
Andrej Karpathy
 @karpathy
21 Jan 2018
Instead trying to describe an architecture in a paper with words, tables and diagrams across 2 sections and 4 pages, it is 90% of the time possible to just paste the 100 lines of code into Appendix A.
4
9
1
28
I love code, and it contains the whole truth. But often, it can’t tell the story. I don’t know how to describe the relationship between these modules in code.
7
6
2
16
In a similar fashion, one might say that DNA contains the whole truth of an organism...
2
6
I like the practice of team members telling each other the 'story of the system' periodically. It's a practice for social and (therefore) technical cohesion.
3
11
22
Thinks... Should architecture decision records be written as a narrative arc?
3
2
8
I definitely remember someone was experimenting with that. Maybe @tastapod?
1
Replying to @mtnygard @natpryce @mfeathers @Grady_Booch @RonJeffries @tastapod
I think it was @boicy

Jan 21, 2018 · 9:58 PM UTC

2
Replying to @stilkov @mtnygard @natpryce @mfeathers @Grady_Booch @RonJeffries
I’ve definitely talked about it, but then it probably was from @tastapod (although I seem to remember @shs96c talking about team shamen a long time back) - I’m still fond of the choose your own adventure thing - ADR’s that link *forwards* :)
4
2
Replying to @stilkov @mtnygard @natpryce @mfeathers @Grady_Booch @RonJeffries @tastapod
The talk I did at YOW! In December was a choose your own adventure narrative in fact (as is the one I’m doing at the thing in Munich, Stefan) - there is something there about ADR’s though, can you capture the (real) options you have *next* to choose from
2
2