Boost logo

Boost :

Subject: [boost] encouraging review managers -- was Re: Review Request: Variadic Macro Data library
From: Gordon Woodhull (gordon_at_[hidden])
Date: 2011-02-19 15:57:48


Hi Ron, Ed, all,

On Feb 18, 2011, at 10:58 AM, Ronald Garcia wrote:
> On Feb 18, 2011, at 10:06 AM, Edward Diener wrote:
>> On 2/18/2011 9:36 AM, Ronald Garcia wrote:
>>> At this time you should try to find a review manager for the library.
>>
>> I am not sure how one goes about doing that, but I will assume I am supposed to ask for one on this mailing list.
>>
>
> That's right. The review wizards will also be putting out a report which requests managers for libraries that don't have one, but individual appeals are often more effective.

I have been mulling over Joachim's Review Manager Assistant proposal, and I have a couple of thoughts about review manager assignments.

First, as Joachim and many others have pointed out, the review "queue" that's described in
http://www.boost.org/community/reviews.html#Wizard
doesn't exist and isn't the right idea. I don't think it makes sense for the Review Wizards to try to assign review managers to libraries - it's much better for people to volunteer based on their own interests and expertise. That is what is happening in practice.

Ed, you volunteered to be a review manager about a month ago. Have you approached any of the authors of prospective libraries on
http://www.boost.org/community/review_schedule.html
who are listed as needing review managers?

I think rather than a queue of review managers, there is really a bag of libraries wanting review managers, and prospective review managers should just volunteer and shouldn't be chosen by the Review Wizards. It would be incredibly difficult for the Wizards to judge whether someone is going to do a good job of managing a review, and even harder to assign a review manager to a library review.

I've only read the list for the past four years so I don't know if the queue was once functional. Wizards serve a valuable role by planning the review schedule and mediating any problems that come up in reviews. But I think that keeping track of prospective review managers without realistically being able to assign them to reviews is needless bureaucracy. The only use for this I can see, is if this list were visible so authors seeking review managers could know who they might write to personally.

I'd like to see the cruft removed from the Formal Review Process. It's confusing and discouraging.

Second, Joachim proposed the role of Review Manager Assistant as a way for new authors like Ed (and myself and many others) to manage reviews in conjunction with more seasoned boosters. The Assistant would do most of the work of summarizing the debate, and then the senior Review Manager would make decisions and produce the final report.

I would like to suggest a generalization which is more fluid (and which would allow RMAs to put the nicer title "Review Manager" on their CVs): simply allow multiple Review Managers for a review. The managers can decide how to split up the work: in sequence like Joachim's RMA/RM idea, or in parallel by divvying up topics for the report. Or some mix of the two.

Having sorted through 260 messages with many dozens of topics when I served as replacement review manager for Constrained Value, I can attest that it's a lot of work to manage a review. Besides making the task more "manageable", having multiple sets of eyes should help ensure that that the report is fair and nothing is missed.

Of course multiple managers would have to agree on the result, but I'll leave off of more difficult subjects for now. (The other being what to do about the possibility of badly managed reviews or disputed reviews, which AFAIK have not been a problem so far.)

On a personal note, I just want to say what a valuable experience it is to manage a Boost review. MPL.Graph's code and documentation are almost ready for review, and I feel that I am also psychologically ready because of what I learned from managing the Constrained Value review last summer. I recommend the experience to all prospective authors.

Gordon

P.S. A lot of discussion on this topic has assumed that a Review Manager has to be an accepted author, but the Formal Review Process just says that they have to be an "an active boost member not connected with the library submission", which I interpret as an active member of the mailing list. Am I correct?

P.P.S. I'd volunteer to manage this review, as I'm eager to be an RM again, but I don't know much about macros.


Boost list run by bdawes at acm.org, gregod at cs.rpi.edu, cpdaniel at pacbell.net, john at johnmaddock.co.uk