Very intriguing but just like most other articles on alternative management structures, this leaves out all the critical information and focuses on praising system without actually understanding it. While I understand that at CERN most people are working for pure passion at modest salaries while being very highly qualified as opposed to someone in 20s with goal to cash out stock options and retire ASAP, I think it's important to figure out if this is the driving reason behind success of loose management structure with no chain of command.
Any alternative management system proposal needs to answer questions like,
1. How people gets hired? Who creates job posting, how interviews are conducted, who does negotiations, approvals and how talent gets attracted and retained?
2. Is there differentiated performance reviews? If so who exactly conducts, signs off these? Is there curve? Is there expected distribution? Who approves promotions/pay raises? Who sets up these rules?
3. If there minimum expectations for performance? Who determines firing and how?
4. If there is no real manager and everything gets decided by commeeties, who sets up these commeeties? How work assignment is done? Who is accountable for tracking progress, success or failure? Who has final say in ties when conflicts occur?
5. What options employees have when they want change? How transfer happen? Who approves these and what are the official rules?
This PR piece is utterly misleading. CERN copied and implemented matrix management, essentially what drove NASA JPL to the ground (ask Ron Garret, lisper). For a hint on NASA's future (not talking of the military) recall the name of the new space telescope. SLAC essentially closed down along with most other labs.
Those who are at the receiving end of the job delegation process report to several managers, none of whom need to take responsibility. Therefore only the one who reports can take hits, as in hightail (see win-win analysis of The Office on ribbonfarm). Obviously success is management's success, not least of the DG.
CERN management proper on the other hand is a tightly knit group (not talking about external contributors, “users” in CERN parlance) with the same level of transparency that need-to-know mechanisms allow for. There are even line-managers and the whole lot. Gianotti was earmarked for leadership position two or three decades ago and has been at CERN since her early twenties. Her external institute affiliation is just a mere formality in the sense that Italy has financially contributed to the CERN budget.
Make no mistake: there is a chain of command, but one that is heavily concealed behind a thick layer of administration (see ribbonfarm) protecting management. There are whole departments and groups devoted to implement the five questions you are wondering about.
I worked on the ATLAS experiment for 6 years. The article gives a reasonable description of the management structure and the unique upsides of that way of working. However it mostly skipped over the negative aspects.
Possibly due to the described desire for consensus, I found the organisation to be incredibly bureaucratic with incredibly lengthy processes. Releasing a paper usually involved around a dozen rounds of review with various groups, often arguing for days about linguistic style more than Physics content.
The lack of clear top-down control makes resource allocation very challenging. There were frequent complaints that Higgs analyses had too much manpower while less "sexy" tasks were chronically understaffed.
The lack of clear assignment of responsibilities also leads to lots of nasty internal politics between institutes. Especially the Higgs analysis where people were eternally engaged in attempts to land grab so they could claim responsibility for bits of the eventual discovery.
Overall, I enjoyed working there a lot. It is a unique structure and the sense of teamwork and lack of hierarchy is very nice. But this article is a bit of a whitewash. I don't think it should be lauded as some incredible model, it has at least as many problems as any other organisation of its size.
JPL uses an organizational structure called matrix management. There are two orthogonal management structures, one organized according to expertise (the "line management") and another, almost completely independent one organized according to task (the "program office"). The program office's job was to win contracts, and the line management's job was to provide the people to work on the resulting projects. Through most of JPL's history, its contracts were large NASA missions with budgets in the billions of dollars, and so this management structure made a certain amount of sense. A vast amount of paperwork had to be generated to win even a single contract.
For most of my career I worked on contracts that had already been awarded, so the inner workings of the program office were completely opaque to me (and still are to this day). By the time I got involved in a project the contract had already been awarded according to a proposal that had been generated by some mysterious process that I never fully understood, and which I never actually saw. All I knew was that my line management gave me a set of account numbers, and I wrote those account numbers down on my time card, and I got paid every other week. And this was true of most of the rank-and-file engineers that I worked with.
The result of this opacity was that the incentives and reward structure for individual employees was often in direct conflict with the goals of the Lab and NASA. For example, one of the factors that went into my performance review every year was how many papers I had published. In fact, this was one of the major considerations because it was one of the few things that management could get a quantitative handle on. So naturally I put a lot of effort into getting published. The problem is that the things you have to do in order to get published are often very different from the things you have to do in order to actually be productive on a NASA project. Getting published requires getting approval from your academic peers, who work for different institutions, often competing for the same contracts that your institution is trying to win. The result is a lot of politics and mutual back-scratching (and back-stabbing), because those are often more effective strategies to get papers published than actually doing worthwhile research.
[...]
At the peak of my JPL career I attained the rank of Principal, which is the highest rung on the technical career ladder whose existence is publicly known. (It turns out there are "secret" promotions you can get after that.) It's essentially the equivalent of getting tenure at a university, only with no teaching responsibilities. The decision to promote someone to Principal is made by a committee. I never found out who was on that committee, or what criteria they used to make the decision. But whoever they were, they had the power to render me more or less un-fireable, despite the fact that by then I was not really contributing anything to the Lab's mission.
I've only worked very briefly in the CMS collaboration, but I've spent a lot of time with people in particle physics collaborations and at medium/large companies. My impression has always been that particle physics collaborations are very well managed for their size, among the best managed human organizations Earth, but that obviously there are friction in getting 3k extremely ambitious scientists to agree on this.
For instance, the "frequent complaints that Higgs analyses had too much manpower" may be valid, but they're not really an example of organizational disfunction unless you had some objective reason to think they were making the wrong decision. Folks disagree about things whether they are inside organizations or not, and if those things are important the disagreement may be venomous. Likewise, the lengthy review process is painful and exhausting, but my impression from almost everyone is that it is sound -- comes to the right conclusions -- and that the conservatism is justified by the field's history of purported discoveries that were later retracted. This soundness is even more impressive when compared to the frequent complaints about the capricious nature of the refereeing process at regular journals, which involves just a handful of physicists. (This is to be compared to physics collaborations outside HEP, like Planck, which are often said to be more dysfunctional.)
So I guess I'm wondering if you could say more comparing ATLAS to similar organizations? "The lack of clear assignment of responsibilities" could certainly be a good criticism, and one that I would actually prefer to apply to physics as a whole. The major rewards are fame and esteem, rather than money, so everyone chases the sexiest, most highly visible tasks.
6 comments
[ 5.6 ms ] story [ 25.0 ms ] threadAny alternative management system proposal needs to answer questions like,
1. How people gets hired? Who creates job posting, how interviews are conducted, who does negotiations, approvals and how talent gets attracted and retained?
2. Is there differentiated performance reviews? If so who exactly conducts, signs off these? Is there curve? Is there expected distribution? Who approves promotions/pay raises? Who sets up these rules?
3. If there minimum expectations for performance? Who determines firing and how?
4. If there is no real manager and everything gets decided by commeeties, who sets up these commeeties? How work assignment is done? Who is accountable for tracking progress, success or failure? Who has final say in ties when conflicts occur?
5. What options employees have when they want change? How transfer happen? Who approves these and what are the official rules?
Those who are at the receiving end of the job delegation process report to several managers, none of whom need to take responsibility. Therefore only the one who reports can take hits, as in hightail (see win-win analysis of The Office on ribbonfarm). Obviously success is management's success, not least of the DG.
CERN management proper on the other hand is a tightly knit group (not talking about external contributors, “users” in CERN parlance) with the same level of transparency that need-to-know mechanisms allow for. There are even line-managers and the whole lot. Gianotti was earmarked for leadership position two or three decades ago and has been at CERN since her early twenties. Her external institute affiliation is just a mere formality in the sense that Italy has financially contributed to the CERN budget.
Make no mistake: there is a chain of command, but one that is heavily concealed behind a thick layer of administration (see ribbonfarm) protecting management. There are whole departments and groups devoted to implement the five questions you are wondering about.
Possibly due to the described desire for consensus, I found the organisation to be incredibly bureaucratic with incredibly lengthy processes. Releasing a paper usually involved around a dozen rounds of review with various groups, often arguing for days about linguistic style more than Physics content.
The lack of clear top-down control makes resource allocation very challenging. There were frequent complaints that Higgs analyses had too much manpower while less "sexy" tasks were chronically understaffed.
The lack of clear assignment of responsibilities also leads to lots of nasty internal politics between institutes. Especially the Higgs analysis where people were eternally engaged in attempts to land grab so they could claim responsibility for bits of the eventual discovery.
Overall, I enjoyed working there a lot. It is a unique structure and the sense of teamwork and lack of hierarchy is very nice. But this article is a bit of a whitewash. I don't think it should be lauded as some incredible model, it has at least as many problems as any other organisation of its size.
For most of my career I worked on contracts that had already been awarded, so the inner workings of the program office were completely opaque to me (and still are to this day). By the time I got involved in a project the contract had already been awarded according to a proposal that had been generated by some mysterious process that I never fully understood, and which I never actually saw. All I knew was that my line management gave me a set of account numbers, and I wrote those account numbers down on my time card, and I got paid every other week. And this was true of most of the rank-and-file engineers that I worked with.
The result of this opacity was that the incentives and reward structure for individual employees was often in direct conflict with the goals of the Lab and NASA. For example, one of the factors that went into my performance review every year was how many papers I had published. In fact, this was one of the major considerations because it was one of the few things that management could get a quantitative handle on. So naturally I put a lot of effort into getting published. The problem is that the things you have to do in order to get published are often very different from the things you have to do in order to actually be productive on a NASA project. Getting published requires getting approval from your academic peers, who work for different institutions, often competing for the same contracts that your institution is trying to win. The result is a lot of politics and mutual back-scratching (and back-stabbing), because those are often more effective strategies to get papers published than actually doing worthwhile research.
[...]
At the peak of my JPL career I attained the rank of Principal, which is the highest rung on the technical career ladder whose existence is publicly known. (It turns out there are "secret" promotions you can get after that.) It's essentially the equivalent of getting tenure at a university, only with no teaching responsibilities. The decision to promote someone to Principal is made by a committee. I never found out who was on that committee, or what criteria they used to make the decision. But whoever they were, they had the power to render me more or less un-fireable, despite the fact that by then I was not really contributing anything to the Lab's mission.
http://blog.rongarret.info/2008/09/failure-must-always-be-op...
For instance, the "frequent complaints that Higgs analyses had too much manpower" may be valid, but they're not really an example of organizational disfunction unless you had some objective reason to think they were making the wrong decision. Folks disagree about things whether they are inside organizations or not, and if those things are important the disagreement may be venomous. Likewise, the lengthy review process is painful and exhausting, but my impression from almost everyone is that it is sound -- comes to the right conclusions -- and that the conservatism is justified by the field's history of purported discoveries that were later retracted. This soundness is even more impressive when compared to the frequent complaints about the capricious nature of the refereeing process at regular journals, which involves just a handful of physicists. (This is to be compared to physics collaborations outside HEP, like Planck, which are often said to be more dysfunctional.)
So I guess I'm wondering if you could say more comparing ATLAS to similar organizations? "The lack of clear assignment of responsibilities" could certainly be a good criticism, and one that I would actually prefer to apply to physics as a whole. The major rewards are fame and esteem, rather than money, so everyone chases the sexiest, most highly visible tasks.