The framework is explicitly titled "Measuring developer productivity with the DX Core 4". Diffs/engineer, even if not at the individual level, is still meant as a proxy for individual throughput. The question they are trying to answer is "how do we make Ganesh X% more productive?"
My argument is that that is the wrong question to be trying to optimize for, regardless of how you measure it. It also does not capture the risk dimensions with AI adoption (and the reduction in humans-in-the-loop) which my paper tries to directly address.
I'm the author – I spent 6 months writing this myself, and, if it helps, our product doesn't even do the stuff in the framework. I'm not sure what I'd be selling?
I know I'm biased but Core 4 (and similar) rub me the wrong way – measuring individual developers as the atomic unit IMO is always meaningless. It's a proxy for organizational health but not directly correlated.
Especially now with AI, what do metrics like "prs/engineer" even mean when you have background agents open/reviewing/releasing PRs without human intervention? what is the right unit for measuring health of the org?
FWIW I write a lot more about "why not existing frameworks" in depth in the full paper.
Initiatives are defined specifically as non-productive, technical leverage-producing initiatives that affect the org's health as a whole and are often left behind. For example, we recently ran an initiative around feature flag cleanups and full rollouts that we tracked religiously during our weekly OpEx review – without which we probably would not have had the same success with that cleanup initiative.
I understand your cynicism with "yet another framework" but (and I know I'm biased) this framework is intimately tied to ops reviews as a mechanism for both measurement and organizational change.
Dev productivity metrics frameworks are (and always have been, IMO) bad for measuring organizational effectiveness. Operational Excellence reviews are the future of measuring and improving organizational health, and I wrote a paper making the case for this + a framework that helps make the OE review effective.
(Shameless plug) – founder of https://www.cortex.io (YC W20) here! We build an off-the-shelf developer portal. Backstage is great if you have lots of time/very niche requirements, but most of our users need a system that "just works" to drive ownership/best practices without adding a ton of overhead
I think you just answered your own question - that sounds quite painful for me to maintain and I'd rather pay Dropbox for the time they save me. + all the added benefits of features they can build with a dedicated organization improving this workflow.
Another founder here. That's exactly how this started - tried to find a solution and didn't find any, and obviously not something most employers are willing to invest time into.
We're still thinking through open core, but on prem is definitely on our road map. Realistically, it won't be available in the near future given that we're still iterating rapidly; once it's in a stable spot we do want to support on prem installations.
Glad to hear you've given up red meat. Not to preach too much here, but why just red meat?
Poultry, for example, is no better. Male chicks are dumped onto conveyor belts and ground alive as they provide no value to the farmers (for meat or eggs).
Dairy farms are also equal to or worse than the factory farms you mentioned.
Most shops in SF have oat milk, if you're located in that area. Oat milk has been a game changer for me. All Blue Bottle locations carry Oatly oat milk.
I'm sorry you're in a situation that makes you feel this way. I've had great experiences with managers that honestly do want to help me - whether that's with advice, giving me opportunities that will help me grow, or even telling me tactical things I should do for promotions (that I've gotten, so it's not just talk with no action).
The framework is explicitly titled "Measuring developer productivity with the DX Core 4". Diffs/engineer, even if not at the individual level, is still meant as a proxy for individual throughput. The question they are trying to answer is "how do we make Ganesh X% more productive?"
My argument is that that is the wrong question to be trying to optimize for, regardless of how you measure it. It also does not capture the risk dimensions with AI adoption (and the reduction in humans-in-the-loop) which my paper tries to directly address.