4 ms·

If you want a shipping culture then encourage, praise, and promote people who actually ship features. If this is a new mindset shift, you will have some crabby

If you want a shipping culture then encourage, praise, and promote people who actually ship features. If this is a new mindset shift, you will have some crabby people who complain about supposed corners being cut, red tape not worshipped properly, and all manner of reasons why the features being shipped are bad. You should listen to the concerns but redirect them to work on automating such corners and red tape and make them see that those getting promoted are actually shipping features and improving the system rather than complaining. Especially in micro services environments the number of commits do not reflect genuine work. Someone can push a boilerplate change every day to one of the 17 repos required to ship and make it seem like they are working when in reality it’s just bureaucracy. Change everything to focus on ease of shipping high quality features and eliminate excuses as to why that can’t be happening.

yourapostasy 4y ago
> ...then encourage, praise, and promote people who actually ship features. If only it were that simple. I've been inside clients where this happened. It generally wasn't pretty. You want an avalanche of trivial features? This is how you get an avalanche of trivial features. Meanwhile, the end user population is restless with non-trivial features sitting in backlogs for quarters and years on end... Then some bright bulb decides to let end users vote which features get worked upon. And the largest cohort of users dominate the agenda driving away the multitude of 1-3% cohorts that happen to make up the majority of the market for the product with their Microsoft Office-like lists of not-quite-alike feature lists, but are the bread and butter of most mature codebases serving large diverse end user populations. Sales people love large diverse end user populations. They make sales so much easier. There is no substitute for judicious leadership. The map is not the territory. You need a balance that you will always struggle to adjust; all metrics or all gut feel both lead to poor outcomes.
- seibelj 4y ago
  It will be on engineering and product management to know what is a trivial vs. non-trivial feature and decide the product roadmap. This is not a question of what is the valuable feature. I’m saying if you have an organization that can’t ship anything, then you need to right the ship by promoting a mindset of actually building and shipping value. The value is not determined by the ICs.

yourapostasy 4y ago

> ...then encourage, praise, and promote people who actually ship features. If only it were that simple. I've been inside clients where this happened. It generally wasn't pretty. You want an avalanche of trivial features? This is how you get an avalanche of trivial features. Meanwhile, the end user population is restless with non-trivial features sitting in backlogs for quarters and years on end... Then some bright bulb decides to let end users vote which features get worked upon. And the largest cohort of users dominate the agenda driving away the multitude of 1-3% cohorts that happen to make up the majority of the market for the product with their Microsoft Office-like lists of not-quite-alike feature lists, but are the bread and butter of most mature codebases serving large diverse end user populations. Sales people love large diverse end user populations. They make sales so much easier. There is no substitute for judicious leadership. The map is not the territory. You need a balance that you will always struggle to adjust; all metrics or all gut feel both lead to poor outcomes.

seibelj 4y ago
It will be on engineering and product management to know what is a trivial vs. non-trivial feature and decide the product roadmap. This is not a question of what is the valuable feature. I’m saying if you have an organization that can’t ship anything, then you need to right the ship by promoting a mindset of actually building and shipping value. The value is not determined by the ICs.