4 ms·
There's a lot of snake oil for sale if you are looking to spend money on solving this problem; this probably is more of that. The reality is that mostly we've
by jillesvangurp 8y ago
There's a lot of snake oil for sale if you are looking to spend money on solving this problem; this probably is more of that.
The reality is that mostly we've gotten better at avoiding things that clearly don't work or are historically obviously misguided/inappropriately expensive (waterfall, CMM level 5, etc); and instead emphasizing things that don't work slightly better: managing risk by doing iterations, not attempting to plan the unplannable eventualities too far ahead of time, etc.
Some people refer to this as Agile. Other people as common sense. Either way, not wasting time on things that clearly don't add value tends to free people up to do something productive (duh). There's a pattern with agile with mostly non technical people higher up the management chain getting overly excited about things like estimates, velocity, burn charts, etc. I usually call stuff like that the illusion of progress and waterfall in disguise. Scrum particularly seems to have devolved to decorating offices with post its and employing busy looking people with moving those around and manually tracking them in convoluted tools like Jira.
But undeniably, we've gotten better over the past decades at building stuff with huge groups of people. Any idiot can probably gobble together some lines of code that does something vaguely useful. But committing to building stuff with hundreds or thousands of people is a different game. It requires lots of money and focus and there are quite a few companies that are doing this successfully.
A bigger pattern in our industry is that people seem to have shifted to calendar driven roadmaps for the most important bits of software where they ship whatever is ready on a fixed dates instead of committing to a long list of stuff shipping whenever it is ready. E.g. Apple ships OS versions once a year, Mozilla, Linux, Chrome, etc. ship every few months, typically with massive amounts of code changes.
- nradov 8y agoCMM level 5 does work, if you actually do it instead of just treating it as a paperwork exercise to satisfy an auditor. There's nothing in CMM that's contrary to or incompatible with Scrum or other agile methodologies. The major emphasis in CMM is on documenting your process (whatever that process is), training people properly, and continuously improving.
- jillesvangurp 8y agoYeah it works. It's just not very appropriate in a lot of situations because it is insanely expensive because it adds a lot of non functional bureaucracy and time consuming activities to the critical path of delivering software.