LaurentB
Laurent Bossavit
Member
Posts: 22
Registered: 21 May 07
Member Is Offline
|
![[*]](./images/grass/default_icon.gif) posted on 24 May 07 at 11:26 AM |
|
|
Proposal: self-certification
I have an idea which could, at one blow, address several of the needs identified in the discussions so far:
* 'advocate for members' by helping them stand out in the job market
* helping practitioners, as well, to hook up with more suitable employers
* provide a concrete solution to the 'verifiable contribution to the community' idea
* provide many of the benefits of certification, without the problems identified in AA's statement on that topic
* reinforce our focus on excellence, on "doing it" as opposed to "talking about doing it".
The idea: "the Agile Alliance comprises a professional network of software development practitioners, who certify each other in the areas of
excellence collectively known as Agile".
The implementation, as I'm currently thinking about it, steals ideas from Advogato, LinkedIn and Twitter.
The basic process is: people sign up to the AA-operated site (let's find a catchy name for it later), after which they can Link with each other or
Certify people they've already Linked with.
* a Link is a statement of acquaintance: I work with X, I've read X's writings, etc.
* a Certification is a much stronger statement. Example: "I, Laurent Bossavit, a member of the agile community, certify Rachel Davies as having
outstanding skills in Retrospective Facilitation, based on the following evidence: I attended a Retrospective which she facilitated at a conference
and took away key learnings from the experience."
(The template for a Certification is roughly the above; it's important that it should include all parts: someone who states that he/she is a member
of the community, the name of someone he/she wants to certify, an area of competence, an assessment of the person's skill in that area : serviceable,
highly valuable, outstanding; and mention of the substantial evidence on which this judgement is based.)
From Advogato we can steal the idea of a trust metric (see e.g. http://www.usemod.com/cgi-bin/mb.pl?TrustMetric ), so that from the overall structure of the network the system derives an
overall assessment: "Rachel Davies is certified by the agile community as a Master in the area of Retrospective Facilitation (details of individual
Certifications given below)".
People who have worked with each other on teams can Certify each other on their programming skills. Team members can Certify their project leader.
Project leaders can Certify their boss (in areas like "Managerial support of Agile teams"). Technical people can certify their Product Owners.
From Twitter we can steal the idea of posting items, like "This is what I've been doing recently in the Agile community (you might want to write me
a Certification if you've found it valuable)". For instance, if you've written an open source testing tool, people can Certify you in the areas of
programming or providing infrastructure support.
The nice feature of such a system is that it's evidence-based; no one has to trust an overall certification, they can look at the details of
individual Certifications.
We may, but need not, distinguish AA members from non-members.
One idea is that signing up to the AA automatically earns you one Certification from the AA in the area of "willing to spend effort or money on
self-improvement". Perhaps extra services as well, more or less on the same model as paying subscriptions to the social networking sites entitle you
to a few extras over the free sign-up.
What do y'all think ?
|
|
|
DaveNicolette
Dave Nicolette
Member
Posts: 4
Registered: 11 Sep 06
Member Is Offline
|
| posted on 24 May 07 at 12:21 PM |
|
|
+1.
|
|
|
simonbaker
Simon Baker
Member
Posts: 19
Registered: 09 Sep 06
Location: London, UK
Member Is Offline
|
| posted on 25 May 07 at 10:35 AM |
|
|
I think this has some merit and is worth exploring more. It's similar to my thinking when I posted a comment on the Agile Practitioners Forum (http://agilepractitionersforum.com/2007/05/01/agile-on-the-march) concerning certification and Jason Gorman's idea of "it
takes one to know one".
|
|
|
debhart
Member
Posts: 72
Registered: 04 Sep 06
Location: Toronto, Ontario
Member Is Offline
|
| posted on 25 May 07 at 07:02 PM |
|
|
+1
|
|
|
PaulOldfield1
Member
Posts: 62
Registered: 05 Sep 06
Location: Perthshire, Scotland
Member Is Offline
|
| posted on 26 May 07 at 11:59 AM |
|
|
Peer certification has always been a favourite of mine. How do we get round the possibility that a group will collude to recommend each other?
Perhaps negative reports "I worked with this person and believe the recommendations are overstated" will also be needed? How does someone working
alone with respect to other agile people get certified? They can still be effective at getting agility across to their team.
Needs a bit of work, but a good idea.
Paul Oldfield
Capgemini, Telford UK
|
|
|
marick
Brian Marick
Member
Posts: 68
Registered: 20 Apr 07
Member Is Offline
|
| posted on 26 May 07 at 12:43 PM |
|
|
| Quote: |
How do we get round the possibility that a group will collude to recommend each other? Perhaps negative reports "I worked with this person and
believe the recommendations are overstated" will also be needed?
|
Back during the certification dispute in the testing community, I wanted to try something similar to Laurent's idea. Colluding groups is actually a
benefit. See, those of us against certification thought it was silly to enshrine one approach to testing as the right way. What the consumer of
certification ought to want is not "this person knows how to do testing right", it's "if you find person X's approach congenial and successful,
this related person Y is probably a good match, but person Z is in a largely separate group and is likely to be a shock."
Besides laziness, the reason I didn't try it is a problem with recommendations: it's easier to give one than refuse one. For a while after that
cursed day when someone convinced me to join LinkedIn, I dutifully turned people down who wanted to link to me if I hadn't worked directly with them.
But it got to be a hassle explaining myself, and it's much easier to click twice and accept, so that's what I usually do now.
I dub this the "Sturgeon effect" for the science fiction writer Theodore Sturgeon, who-although the person who coined the phrase "95% of everything
is crap"-was notorious for blurbing any book he was asked to. The insiders knew this, so knew that a Sturgeon blurb on a book meant nothing. But the
casual reader-the one supposedly served by blurbs-took the blurb seriously.
The penalty for giving a recommendation that's advertised as meaning "I vouch for this person" but actually means less is distant, unlikely, and
probably small. There needs to be better feedback for recommendations. Perhaps making a recommendation more of a letter of recommendation, as Laurent
recommends, is the way. And I expect the social software people may have solved this problem. So I'd support a program to this end.
|
|
|
mfeathers
Michael Feathers
Member
Posts: 15
Registered: 22 May 07
Member Is Offline
|
| posted on 26 May 07 at 10:18 PM |
|
|
One of the things that bugs me about certification is that it is a cop-out. Effectively, it's a way of outsourcing the development of your hiring
criteria.
Organizations have trouble hiring above themselves. What they really need, in many cases, is to hire someone who has the insight needed to hire
others in an area (like agile). They need one tall tree.
When I've seen organizations do well with agile over the long term, I've noticed that their tallest trees have gotten involved in hiring.
I guess that certification and self-certification could help an organization get that first couple of tall trees, but I shudder to think of what the
industry impact would be if organizations started to rely too heavily on it.
Hiring one of those tall trees isn't that much of an investment if you are serious about agile, software development, etc. The thing that I fear
about certification is that it makes it easy to take things less than seriously.
|
|
|
LaurentB
Laurent Bossavit
Member
Posts: 22
Registered: 21 May 07
Member Is Offline
|
| posted on 27 May 07 at 12:52 PM |
|
|
Paul: in the scheme I have in mind, you can certify someone whenever you have hard evidence that they know what they are doing in this or that area of
Agile. Someone who works in isolation and wants to be certified will have to show their work to someone who's plugged into the larger community: I'd
consider that a feature. They can seek out someone who's certified to come and assess their project; perhaps the AA can facilitate that process.
Collusion is not a problem if the "trust metric" used for certification has a fixed set of "seeds", people who are axiomatically Agile experts,
who certify others who in turn certify others and so on. But there are other algorithms without that property.
The algortihms are supposed to be somewhat attack-resistant; though we'd need to take that with a grain of salt. I suppose there's no automated way
of fully guarding against "bad certifications" of various sorts. But if everything is transparent - which algorithm is used, who is trusted
axiomatically, what the individual certification statements are - I expect that social norms will make the whole thing robust enough. We can set up a
process whereby certification statements can become subject to review.
|
|
|
PaulOldfield1
Member
Posts: 62
Registered: 05 Sep 06
Location: Perthshire, Scotland
Member Is Offline
|
| posted on 27 May 07 at 01:06 PM |
|
|
| Quote: | Originally posted by LaurentB
Paul: in the scheme I have in mind, you can certify someone whenever you have hard evidence that they know what they are doing in this or that area of
Agile. Someone who works in isolation and wants to be certified will have to show their work to someone who's plugged into the larger community: I'd
consider that a feature. They can seek out someone who's certified to come and assess their project; perhaps the AA can facilitate that process.
|
Okay. Hmm... suppose their work is as an agent of change rather than on a project?
| Quote: |
Collusion is not a problem if the "trust metric" used for certification has a fixed set of "seeds", people who are axiomatically Agile experts,
who certify others who in turn certify others and so on. But there are other algorithms without that property. |
Fine. Not sure about Ken's CSM, but in general, fine. If Ron says anyone is agile, I'll take his word for it. If 10 of us lesser mortals say so,
I'll probably take their word for it provided I've heard one or two of the names around the circuit...
Paul Oldfield
Capgemini, Telford UK
|
|
|
kevinrutherford
Kevin Rutherford
Member
Posts: 2
Registered: 05 Sep 06
Location: Macclesfield, UK
Member Is Offline
|
| posted on 03 Jun 07 at 03:48 PM |
|
|
Great idea, Laurent! And I do like the idea of the AA helping to find someone to parachute in and do an assessment.
Perhaps recommendations should have increased verity when the two parties work in / for different organisations...?
|
|
|
ChrisMatts
Chris Matts
Member
Posts: 20
Registered: 23 May 07
Member Is Offline
|
| posted on 03 Jun 07 at 06:52 PM |
|
|
Sorry to be a kill joy on this one.
I have done a LOT of recruitment as part of my job. You would be horrified at the number of people who lie on their CVs today. The percentage of
people is much higher than you would think. Both young and old. The stakes are much higher in IT these days because it is such a lucrative
industry.
Problem 1. People can create fictitious accounts which all accredit each other. Who validates the site? What if ten people got together and created a
fictious set of people each. People, especially those who know they are not that good, WILL game the system... And they will be smart about it.
Problem 2. Recommendations can only be based on personal experience working alongside someone. Even then, most people will only recommend someone
based on context. "Worked well with ten people they already knew. Not sure how they will get on with other people." As such, there isn't really an
upside recommendation. People will only say, well they weren't that bad when I worked with them. I know Andy Pols, Rachel Davies, Steve Freeman
really well. I would be very cautious about recommending them carte blanche on an accreditation web-site. I would recommend them in a personal
off-line conversation.
Problem 3. When most managers are hiring, they are really interested in avoiding the bad ones rather than getting the good ones. Namely, they want to
see the bad recommendations. Where did this person not work. Apparently, because I've never done it, These conversations tend to happen off the map.
A manager will look at a CV and use their own network to check out a candidate. I cannot imagine the site would last too long if people were
encouraged to publish the negative. This has changed. In the early days of IT, there were few technologies and it was fairly easy to keep track of
them. Post Dot-Com there are lots of new technologies and a lot of lightly skilled developers, often with lots of accreditations.
Finally. In my industry, we tend to only care about real qualifications (rather than certification) and experience. The only real accreditations worth
anything are those for the lawyers, doctors and accountants. They take years to achieve and the accreditation processes have taken centuries to
achieve.
Lets ask a simple question. "Are we trying to fix an immediate problem, such as, "How do the Agile early adoptors differentiate themselves from the
early majority?" or do we want to start on a path for "software" developers in ten, fifty, or even a hundred years?
I do like the Agile version of the Kevin Bacon game. I have a Rachel Davies number of 1 because I once bought her a pint at XTC. :-)
|
|
|
duncan_pierce
Duncan Pierce
Member
Posts: 6
Registered: 03 Dec 06
Member Is Offline
|
| posted on 08 Jun 07 at 03:52 PM |
|
|
| Quote: | Originally posted by mfeathers
Organizations have trouble hiring above themselves. [...] They need one tall tree.
|
Some organizations can't see the tall trees for the dead wood. 
Duncan
|
|
|
RSS Feed
