Mark, Without going into any more detail on your email, I just want to say that your comparison with the percentage needed to form a recall of a sitting British member of parliament is wrong. In that case, the percentage needed is to effect the recall, not to form a petition that is then considered by a committee. The two cases do not stand comparison. I think you raise an interesting point about "whether we want our leadership concerned about the possibility of a recall -- even if a failed one -- from a small, determined set of people. I imagine there are upsides and downsides to that." Again, of course, the recall is not achieved by the small set of people: they achieve the petition, and the recall committee is the group who may or may not achieve the recall. Thanks, Adrian -----Original Message----- From: ietf <ietf-bounces@xxxxxxxx> On Behalf Of Mark Nottingham Sent: 22 April 2019 03:20 To: IETF <ietf@xxxxxxxx> Cc: S Moonesamy <sm+ietf@xxxxxxxxxxxx> Subject: draft-moonesamy-recall-rev-01: Number of Signatures Required Hi SM, Without going into the other aspects of this draft (I'm watching with interest), Section 2.3 caught my eye: > [RFC7437] requires at least 20 signatories for a recall petition with > no more than two of the signatories having the same primary > affiliation. That sets a very high barrier for a recall petition > even though the recall petition requires a, justification, an > investigation by a Recall Committee and a 3/4 majority of the members > of the Recall Committee who vote on the recall decision. This > document also proposes to decrease the number of signatures required > to avoid making it impractical to invoke the first step of the recall > procedures. Sorry if I missed a previous discussion, but this strikes me as an interesting change to couple to a proposed increase in the number of people eligible to sign a recall petition. Given that the number of people who attend IETF meetings -- both physically and remotely -- varies quite a bit, and since we now can track the size of both populations, I'd expect this number to be expressed as a percentage of the total eligible population (however the above discussion goes). Looking at the attendance of the last five IETFs*, it appears that about 777 people are nomcom eligible, so it would currently take a bit more than 2.5% of that community to initiate a recall. If remote participation is also counted, about 1083 would currently be qualified; your proposed 10 signatories is 0.92% of that community. What's the right number**? I'm not sure, but I'll make a few observations. It seems to me that the number of signatories required helps to ensure that the recall represents a significant portion of the community, so as not to waste the noncom's time. I also see it as a way to manage secondary effects; e.g., so that people aren't dissuaded from putting their hands up for nomcom, lest they be committing to a series of frivolous recall efforts. More importantly, perhaps, I think we also need to ask ourselves whether we want our leadership concerned about the possibility of a recall -- even if a failed one -- from a small, determined set of people. I imagine there are upsides and downsides to that. To me, allowing 1% of the community to initiate such an action seems like a really low bar for disruption, not at all a "very high barrier". Looking elsewhere: * Various Swiss cantons seem to run between 2% and 13% of the electorate <https://en.wikipedia.org/wiki/Recall_election#Switzerland> * In the UK, it takes 10% <https://en.wikipedia.org/wiki/Recall_of_MPs_Act_2015> Cheers, * This is estimated by scraping the attendance tables for the last five IETFs, removing affiliation (since it might change) but keeping country (less likely to change, but still possible), case-normalising and doing some command line sorting and filtering. No doubt mistakes were made (if anything, these numbers are probably low, thanks to false negatives), and I'm sure this isn't how the actual eligibility is determined. But it's good enough for estimating, I think. ** I really do think it should be a percentage. -- Mark Nottingham https://www.mnot.net/
