I’m coming back to the session Agile & CMMi: magic potion or great divide that we recently gave, Yassine (@yassinezakaria) and myself, at Agile Tour Toulouse 2011. The slides are available here (or embedded below) and I wanted to comment on them a bit, even though written comments rarely replace a session.
First, you should know that this session led Yassine & me to quite a few rather heated discussions.
Waterfall idea:
Why did we do a session on Agile & CMMi? First because we find the rallying cries and prejudices that a large part of the people from these two worlds display towards each other a bit absurd. We had fun putting this into form through different slides. But the unfortunate reality is that in people’s minds these caricatures are not far from being true. And moreover, many poor applications of agile or CMMi often make them true or reinforce them…
These prejudices are all the more damaging because:
- whether it’s agile or CMMi, we often refer to the “chaos report” from the Standish Group to explain that we want to put an end to failing projects, as evidenced by our different training slides…identical ones.
- the initial response takes a comparable form: a set of practices on which we operate a focus: integrated set for CMMi, practices pushed to the extreme for XP for example. I specify clearly: an “initial” response, which I didn’t do in the presentation or the slides and it’s still today a subject of discussion (over hors d’oeuvres) with Thierry (@thierrycros).
These prejudices are all the more unfortunate because our “great experts” (some “big names”) are seeking to operate a convergence between these two approaches (CMMi & Agile). We give as evidence some feedback from Jeff Sutherland, David Anderson, Richard Basque or Scott Ambler. On this subject Tullius Detritus takes pleasure in reminding you that you shouldn’t necessarily believe the “great experts” blindly and that a little perspective never hurts. (In this context Jeff Sutherland talks about “magic potion”, which is what gave its name to this session, or “great divide” we can wonder).
We then present the different ways in which each of us on our side can benefit our approach from certain qualities of the other. In a way quite consistent with the approaches of Sutherland or Basque, Yassine tries to have his CMMi projects benefit from certain good agile practices: notably feedback (often too late in CMMi), iterative aspects, etc. It should be remembered that CMMi naturally proposes: stakeholder involvement, making the necessary adjustments to each project (we don’t embark on a unique and monolithic CMMi).
[added Oct 29] The classic approach consists of saying that CMMi is the “what”, and agile the “how”. As an agilist I want to change this paradigm. Agile is fundamentally at the center of the reflection because it is a culture, it brings values. So schematically it is the “why”, upstream of the “what” and the “how”.
For my part, I quite agree with David Anderson on the fact that Agile could benefit from level 4 & 5 practices of CMMi, either to rely on a framework to manage innovation at the organizational level, or to rely on trends to better manage projects. Be careful however, metrics for metrics’ sake is naturally a danger. QPM level 4 of CMMi doesn’t actually propose relying on numbers to manage projects, but on trends. This practice that we would draw inspiration from just tells us: here are the statistics that we were able to collect, be attentive if your project seems not to conform to them when it should. Just be attentive, aware (forewarned is forearmed) and not necessarily more. For these statistics to work (that’s also a word that I don’t like too much, and that must be handled with care), they must come from information that can be comparable between projects. And therefore naturally you can already exclude velocity. What metrics? I can refer you to for example Jérémie’s article (@jgrodziski), we could also base it on turnover rate (obtained from HR), etc. Be really careful therefore not to alienate yourself with numbers and to think carefully about indicators (difficult to use business value, unthinkable to use velocity, etc.)
I’ll briefly discuss in a future post the “CMMi fortunes” that I propose. It’s about relying on the enormous retrospective on which CMMi is based to -in a playful, simple and regular way- level up agile teams that might be lacking “experience”.
Finally, our conclusion wants to convey two messages:
- Whatever your choice (Agile or CMMi) you must give yourself the means for your approach, no point in pretending, only for marketing purposes, for certification, etc. You really must try to give yourself the means to truly embrace the approach.
- We most definitely do NOT recommend a hodgepodge of these two methods. You must have a clear and unambiguous backbone: agile OR cmmi. Then why not enrich yourself with contributions from other methods, that’s our point of view.
Someone asked me a question: how do you know that you’re not doing “marketing” agile. The answer seemed so obvious to me that… I didn’t have an answer. This will also be the subject of a future post. How to be sure that you’re not doing “marketing” agile. Thanks to the person who asked this question that remained without a real answer.
As illustration at the top left and to relaunch the discussion once again, “Waterfall” mode interpreted by the Goths in Astérix & Obélix. I don’t know if there’s a connection but this post was written with “Overkill” by Mötorhead in my ears. I don’t come out of it enhanced.
Agile & CMMi, magic potion or great divide?