01 December 2007

CMMI is overrated and unnecessary...

This entry over at OpenSource Connections got me thinking.

They're absolutely right.

I've even said as much myself at NDIA's 6th CMMI Technology Conference in 2006 in response to a question.

The question was something like, Are you saying that CMMI isn't necessary if our company believes we are fine the way we are, and we don't see any need to improve?

My answer was pretty much, "yes".  I added, if a company is OK with the status quo and competition isn't a concern, and their leadership doesn't believe in the business rule-of-thumb about not moving forward is the same as falling behind... then yes, CMMI is overrated and unnecessary...

... for companies who don't worry about lottery-sensitive individuals, or rapid growth, or shifting products, or who are quite content with the strength of their client relationships that they need not worry anyone will step in and oust them or bring in competition.

I'm always thrilled when I come upon a software company that has its act together like OpenSource Connections seems to have.  However, as regular readers know, I'm not as thrilled that the folks over there have an entirely inaccurate perception of CMMI as the ability "to check certain boxes about whether or not we have a laundry list of processes documented and implemented."

CMMI's language certainly makes one believe the assumption in the model is that people aren't doing the right things.  However, the exact wrong way to view or use CMMI is as a list of check-boxes.  Anyone found performing an appraisal in that way ought to be reported to the SEI.

But that's not the point of this post.  The point here is that you should read the OpenSource Connections post.  Even with their misunderstandings they make good points.

One point about the reality of level-rated organizations (even high-maturity ones) can still deliver junk is right on the money.  Not having done an in-depth study myself -- not sure anyone has -- I'd venture that most level-rated organizations delivering junk took the check-box approach to CMMI, whereas the ones who took a business-value approach to CMMI actually deliver customer delight and high profits.

Another point about process having a place in agility was also crucial to giving their entry credibility.  Recognizing that even small, agile organizations do have processes and rely on them for success is a think of beauty.  More organizations would be successful if they understood this much about themselves.

Their shortcomings in understanding CMMI are excusable, the points I wish to surface have to do with a few aspects they didn't address.

One thing CMMI helps avoid (but, by no means is the only solution for) are issues with know-how that's locked in individuals' heads.  Or for dealing with a customer-base that expects specific people to be doing the work.  Even with this CMMI isn't required, and certainly not an entire maturity-level rating, but if an organization is facing these issues and doesn't have any idea what to do about it, parts of CMMI can help.

Which brings me to my second point for the moment.  Maturity Levels aren't the only way to use CMMI.  In fact, even with capable organizations, many find that being able to borrow practices and goals from particular process areas may help them fine-tune an already working operation.

Whether an organization bothers to pursue even a Capability Level for any process area(s) isn't even where I'm going.  Being "rated" has nothing to do with being capable or mature at development.  CMMI has value whether or not organizations choose to get rated; whether or not they choose to use entire maturity levels, entire capability levels, entire process areas, or even entire goals.  CMMI can add value one practice at a time.

Organizations that have their act together could probably do very well with an appraisal and learn a lot about themselves.  Limiting the value of CMMI to whether or not the organization has had an appraisal limits more than CMMI; it limits the business value of creating the efficiencies, knowledge longevity, and ROI for looking into their own processes at all.

Labels:

8 Comments:

At 02 December, 2007 15:12 , Blogger Steven M Smith said...

Hillel, Good for you! Rarely do I hear someone with CMM experience say that the CMMI is unnecssary when an organization's "status quo and competition isn't a concern, and their leadership doesn't believe in the business rule-of-thumb about not moving forward is the same as falling behind."

I would like to simplify your statement further. The CMMI may produce zero value for an organization that is producing the results it needs.

Rather than focus on things the CMMI model considers are important, the organization might produce better results by continuing to tweek their own model.

I think you would agree with me that the CMMI isn't the only model that can help an organization produce better results. At least, I hope you would.

Best regards,
-Steve

 
At 02 December, 2007 16:00 , Anonymous Anonymous said...

It is clear there is no magic and any process compliance only guarantees repeatability of process, stating little about the product. That said your article over simplifies the issues of implementing SQA\SQC with an abstact medium and reduces the serious search for answers to tabloid type headings and discourse. I see no serious arguments here that have positive practical consequences.

 
At 02 December, 2007 19:26 , Blogger Hillel said...

To Steve:

Yes, I would agree that CMMI isn't the only model for improving results. In fact, CMMI explicitly avoids dealing with two other very important aspects of organizations: people and technology.

Other than in the OID process area, there's little in CMMI that explicitly bleeds over into other areas of an organization besides its processes.

Beyond that, even when focusing only on improving processes, CMMI isn't the only thing out there.

Either way, while CMMI can be used holistically within the domain of processes, there are other approaches that are even more holistic across more than just the process domain, any of which -- when applied properly -- can achieve better results. Especially when considering the speed of change as an important attribute.

To Anonymous:
I don't understand your comment. It neither provides anything constructive nor expresses what's actually wrong with my post.

What it *does* however provide is a clue that you do not understand CMMI... having reduced it to SQA/SQC yourself... to which CMMI is not limited. It also sheds light on the fact that you've read none of the other posts in this blog.

If you're interested in serious arguments with positive practical consequences, perhaps you can start by contributing some yourself.

-->> HG

 
At 12 December, 2007 19:42 , Anonymous Anonymous said...

Hillel,

Thank you for commenting on my article about CMMI. Having just come from an AFEI conference focusing on opening up the Department of Defense (open architecture, open source, open processes), I stick by my assertions that dependence on CMMI is not a magic bullet (though, in retrospect, I'd probably calm the tone of my first paragraph). Look at any random sampling of postings on the FedBizOpps website (http://www.fbo.gov) that fall under software development (such as NAICS 541511 or industry category D), and you'll see quite a preponderance of postings where a qualifier for doing work is having CMMI level 3 or higher certification; yet, project managers provide plenty of anecdotal evidence to show that such a dependence is not working like it should. I daresay that past performance would show an unacceptable correlation coefficient between success and CMMI certification.

With that assertion, I think that your next to last paragraph is well articulated. There are many paths to a better organization, and CMMI certainly provides a way. However, blind reliance on a CMMI certification by the purchasers of software development services is an abrogation of responsibility by the purchasing authority.

CMMI does provide a certain standard of certification for those who cannot independently certify the capability of a provider. Just as ISO, Good Housekeeping, etc. provide certain seals of approval, so does CMMI for a buyer. I understand the value of that certification, but believe that blind reliance on it leads to trouble.

I think if you and I sat down over a beer, we'd probably agree on 80% of things, and I appreciate you bringing up in a thoughtful manner the ways in which we do not.

Cheers!
Jason Hull

 
At 12 December, 2007 19:56 , Blogger Hillel said...

Jason,

Actually, the DOD has already come out and said that their contracting officers' collective zeal to require "level ratings" as precursors to bids was ill-conceived from both a theoretical and demonstrable basis. They were told for a long time not to make level ratings a target of achievement. And, go figure, their own data now shows a negative correlation between rated levels and project outcomes!

This comes as no surprise to true process professionals (a minority of the population even among those in my position) who, for a long time, were railing against the focus on level ratings as a chinning bar to getting contracts. Ratings as designed by the SEI were supposed to be used as internal benchmarks of progress, not external assurances of performance.

So, yes, there are probably more DOD contractors using CMMI for pathological box-checking than for actual improvement. Furthermore, if we allow "mob rules" to apply, then the way CMMI has been implemented by the majority would imply that this is how to implement CMMI. As wrong as they might be. This is the "norm".

Very sad.

Thanks for writing in!
-->> Hillel

 
At 23 January, 2008 12:11 , Anonymous Anonymous said...

Hillel , I will certainly agree with you. I have implemented the processes in my organization and the CMMI , the whole SCAMPI A appraisals and all the process things are overheads.

I work in a 3rd world country , and we hear every day that indian companies getting the CMMI level 5 appraisals.

Beside everything inside is dirty, by implementing CMMI its just marketing tactics.

I have worked with lot of consultants and the worst thing i see is people just for the sake of the CMMI drop their natural processes and adopt processes of some company in US or west which has already been appraised. in doing that they lost their own touch and start working in an alienated mode.

rather now if you go and really see Six Sigma and try to improve your own internal processes that you have been originally working with its a much better way. and i always agree with that.

beside by following CMMI - let me tell you the real facts that i have seen

Productivity reduces by 30-40%
Overhead increases by 40-60%
Quality Reduces by 20%

now i would say a person should try to improve in the way he is doing work and rather than getting the Capability Maturity Level and doing what is strictly unneccessary

a very nice post

 
At 26 April, 2008 00:09 , Anonymous Anonymous said...

I would like to know where you obtain the CMMI to performance correlation numbers. We are evaluating adoption of CMMI but feel more inclined to focus on frameworks that will work best for our organization and drive continuous improvement. I have to beleive that govt buyers will sooon start to value and organizations efforts to do what 'best' for them, and ultimately thier client, rather thanchecking the box with CMMI.

 
At 26 April, 2008 21:28 , Blogger Hillel said...

Dear Anonymous,

What do you mean by "the CMMI to performance correlation numbers" ?

If you're asking for data that shows how CMMI has improved organization's performance, those numbers are nearly impossible to collect for the universe of users, even harder to make meaningful so that what would be collected is normalized, and the still harder to make more than anything other than anecdotal.

There's plenty of anecdotal "data", but again, it's fodder for discussion, not really strongly supported.

For one thing, the data will be very different from organization to organization as a result of the differences in states of organizations' situations when they started using CMMI. Expand that thought and you begin to see why even the reported benefits can't hardly be seen as anything meaningful.

Here's an analogy: 18 weeks in the gym, 6 days a week, doing a very specific (and planned) routine every week, consistently increasing my intensity weekly or daily, and not being stupid with what I eat *will* have a predictable result. If I do everything "correctly" it will be a very beneficial result. But that's me. Someone else will have a different result. The options range from having a very negative result to achieving the same results in less or more time.

When it's all about checking the box, people perceive that the approaches and results are comparable. They're not. And, they're *still* not whether or not you're just checking the box.

 

Post a Comment

Subscribe to Post Comments [Atom]

<< Home