Incumbent bias in elections / impose term limits?

[marcoscaceres]

As was clearly evident from the 2022/2023 special election, there is a strong incumbent bias in the election process that should be addressed by the AB with some expediency.

The AB should really put term limits in place (e.g., you can only serve 5 years total on the AB), otherwise we just end up with the same people getting re-elected over an over again - even after they've taken a break from the AB.

That’s not the say the re-elected folks are not great (and you all know that I deeply respect all of you), but it does mean no new ideas/people come through so we just end up with the same ideas/people over and over again. In the long term, that's not a good thing for the community/w3c.

It also means limited chance for new people to get trained up in, or elected to, the AB. So, maybe consider reserving one or two slots for new people. Excluding myself here and meaning no disrespect to the re-elected AB members, looking at the great list of new potential candidates that ran in the last election, it's extremely sad and such a missed opportunity for our community that none were elected.

This should probably be a separate issue, but there is also a significant gender imbalance on the AB. It would be great if the AB could consider putting in gender quotas or some other system that ends up with a make up of genders that is more reflective of society (and not reflective of the the tech industry, where gender equality is extremely bad/harmful). We should be aiming for better and should do more to get there. Again, in the recent election, we had almost had a fairly equitable gender split. It's a shame that that was not reflected in the end result.

[dwsinger]

I think we should address incumbent bias, but I don't think term limits are a good tool. I am also sad that we didn't elect new people. Both the organization and the candidates can help with incumbent bias. I suspect it's to a large extent a question of visibility, so actions like

• making the candidate fora, questions, and so on, more visible, planned in advance, better timed
• if you are a candidate, get involved in governance and be visible; for AC reps, be visible in AC discussions, pro-active in raising questions, helpful in resolving them; get involved in governance – the process CG is very lightly attended, for example

[chaals]

I suspect incumbent bias is largely based on perceptions of the electors that this is a hard job that takes time to learn, and it's important to put people in who can "hit the ground running" in the current climate (unfortunately, this seems to hold for almost any values of "current climate").

Yes, it is a hard job, and historically there is as often a shortage of willing candidates as there are a surplus of qualified ones.

I do think we should be clear that people can learn the particularities "on the job". And a lot of what is required is not unique - it's a question of being able to think, to accurately represent what other people say they want, to organise moderately well and above all being willing to take on actual work and get it done.

I'm not a fan of term limits either, but I don't mind setting quotas. (They have drawbacks as well as benefits, but they are better than just hoping we get some specific result we voters want, if we agree it matters more to us than getting the sum of who we want). I am mildly in favour of a gender quota and could readily live with a small minimum quota for people who have and have not been on the AB before (e.g. at least 1 of each). More particularly if we didn't stagger elections #688 it seems easier to get a better range of candidates elected.

[cwilso]

First and foremost, I'm not convinced we have had problems with incumbent bias long-term. On the whole, we tend to turn over at least one or two members every election in the AB and the TAG; and we have also worked very hard to encourage people to participate in the AB's work even if they are not on the AB (and I think doing so greatly prepares one to win a seat on, and serve on, the AB). I will even more strongly recommend David's advice: if you're interested in being on the AB, participate in the Process CG; poke your nose into the Vision work, et al. I also strongly recommend that if you want to get elected, you do need to participate in campaigning, and reach out to members for their votes. First-place votes matter a lot in crowded elections.

As to gender disparity, I would strongly suggest splitting this into a separate issue (and probably over in ab-memberonly) if you want to further that cause. FWIW, we've had radically better gender balance on the AB than across the Membership (let alone across the tech industry) for some time; that doesn't (of course) mean "we're done", but I don't think it's wise to impose quotas. I do think it is important to both ensure the elected bodies are inclusive to everyone, and to reach out to great potential candidates that will expand our diversity (both of these, fwiw, are things that a number of us on the AB and the TAG actively do). We consistently ramped up the women on the AB from the time I first joined (2013) from 1 to 4. (Briefly, the AB was close to 50% women, in 2018, when it was 4/9.) Although we went down by one in this special election, I have no doubt it will recover.

I'm not entirely sure how we would impose such a quota - or quotas on "have/have not been on the AB before" - other that literally skipping candidates who won the support of the electorate, which seems like a pretty bespoke voting process.

[michaelchampion]

Over the 20 years or so I've been paying attention to AB and TAG elections, it's been pretty clear there IS an incumbent bias. I've argued for term limits in the past, but that has never got much support, mostly for the reasons mentioned in this thread. If I had benevolent dictator powers I'd set a term limit of 8 -10 years, but I've given up trying to build consensus for that.

Going forward I'd agree with @dwsinger and @cwilso -- emphasize efforts to bring in new people and give them visibility, over efforts to exclude incumbents. There are AB-driven efforts that largely work in public view, such as the groups maintaining the Process and CEPC, that are opportunities to get things done and show leadership/build consensus. The AB has started the tradition of monthly membership meetings where people who want to be on the AB and/or have issues with AB actions (or lack thereof ;-) ) could question AB members directly, make the case for different decisions, etc. All these are opportunities for people to become known, and build credibility / constituencies as people willing and able to tackle the AB-level challenge.

Also the Team+AB has been soliciting ideas for the AC meetings in GitHub, and TPAC breakout sessions are mostly self-organized. Bring ideas about what needs fixing and how YOU would do things differently to AC and TPAC sessions to get visibility and build communities to advance those ideas.

I must admit there are people who have run in the recent AB and Board elections who did many of these things, apparently got high rankings from the AC reps I confer with, but still didn't get enough votes to be elected. I feel their frustration. But I also recall that I lost the first two times I ran for the AB, and got in on the third attempt because there were just as many nominees as open seats, so everybody won. I guess after that I benefitted from the incumbent bias 🤷🏽‍♂️. So don't stop running, but do things to demonstrate your vision for a better web, willingness to work hard for it, and ability to build consensus on a way forward. I feel confident that the people who demonstrate to the AC their ability and leadership, will sooner or later win elections.

And remember, the open web community is NOT a hierarchy with the AB entrenched in power and one has to be in the in-group with an authoritative position to get things done. There are many more challenges than the AB elected members can take on, and much more work to do on their priority projects than they can accomplish in the short run. My mentor in standards engagement had a saying "those who do the work, win." In other words, W3C (like most volunteer-run organizations) is mostly a DO-ocracy https://communitywiki.org/wiki/DoOcracy; finding things that need doing, and doing them better/faster than the people who are nominally responsible for doing them, is usually effective. Eventually the larger community (and hopefully W3C's electorate) will recognize the value of your work.

[chaals]

@cwilso said

I'm not entirely sure how we would impose such a quota - or quotas on "have/have not been on the AB before" - other that literally skipping candidates who won the support of the electorate, which seems like a pretty bespoke voting process.

It is a bespoke process, and that is indeed how I imagine we would do it if we decided to use quotas. In practical terms, the biggest complication is getting enough candidates so we don't have a problem doing this, and deciding on the specific mechanism for choosing (see also #694 especially https://github.com/w3c/w3process/issues/694#issuecomment-1374642043 where @cwilso makes the point that many STV algorithms are not obviously designed to rank all candidates. Although Condorcet methods like Shulze_STV do).

[marcoscaceres]

I'm a proponent of, believer in, and part-taker in the W3C's DO-ocracy. It's why I edit more specs than I should, Chair a WG, and continue to make sure spec tooling gets Process updates, bug fixes, etc. etc. @michaelchampion pointed out elsewhere, contributing to the AB takes a lot of time and commitment. And to @chaals' point, contributing to the AB is not particularly difficult... it just takes a long time to get onboarded and keep up with discussions, etc.

It can be hard for someone like me personally to volunteer time to the Process CG because it would mean diverting time from chairing, editing, and implementing, and would require doing so without being formally recognized by the AB. I know asking folks to either volunteer and/or prove themselves in the Process CG comes from a good place. But, at the same time, I ask that folks here be mindful of asking people to volunteer their time because... well, there are only so many hours in the day... and people have day jobs, and getting time allocated from that day job may sometimes require being elected first.

For some, the intent of running for the AB is so one can get (paid) time to work on AB things and to contribute to the CG (by having to perhaps drop other things, by virtue of being elected, which comes with explicit time commitments and recognition).

Maybe a different solution: It the AB is still not willing to entertain term limits, I believe the TAG has (or had?) a means for the Director to appoint one of the positions. Maybe the AB could use a similar mechanism to get new folks on here? That could help solve for some of the incumbent bias, and give upcoming folks the time, recognition, and space to participate.

[dwsinger]

@marcoscaceres The counterpoint to what you're saying is that handling the process, being involved in AC Forum discussions that relate to governance, and so on, are part of being on the AB. If you were to get elected, you'd be expected to do these things, so if you don't have time, it's not clear why you'd run.

The TAG has appointed seats, yes; one theory is that they exist to ensure that the TAG has the right mix of technical skills (security, back-end, user-facing, browser, etc.). I am not sure that the AB has the same issue. Note that we are introducing TAG appointed seat term limits.

[marcoscaceres]

it's not clear why you'd run.

Sorry if I wasn't clear. I'm was trying to say the same things you said above: once elected to the AB, then that elected individual MUST make time for "handling the process, being involved in AC Forum discussions that relate to governance, and so on".

Unelected individuals, in most cases, won't have the time to do the things listed above - and I know no one on the AB is asking of that from unelected individuals, but my point stands.

So, to @michaelchampion's suggestion:

The AB has started the tradition of monthly membership meetings where people who want to be on the AB and/or have issues with AB actions (or lack thereof ;-) ) could question AB members directly, make the case for different decisions, etc. All these are opportunities for people to become known, and build credibility / constituencies as people willing and able to tackle the AB-level challenge.

Sure, but people contribute in different ways - and how one participates might not match a traditional contribution expectation (e.g., implementing the Process in ReSpec and trying to introduce it in a Working Group): As #589 illustrates, and this email shows, the AB has had blindspots about how changes to the Process impacts systems, tooling, and Editors directly - or at least, the AB has left it to others to then deal with any fallout from changes to the Process document.

Those Process changes impact me and my editors directly, and due to limited coordination with systems and tooling, it impacts the W3C community as whole. These are things that I'd definitely give feedback on and a personal motivator for running for the AB: things we can learn from and do better in the future from a project management perspective.

But the feedback is usually given at implementation time (again #589). That could be handled better by the AB engaging with Editors, Chairs, etc. and not expecting chairs, editors, etc. to monitor this repo or the AC list.

See what I mean? (to be clear: be great if the AB went to the editors, chairs, etc. more directly - don't make or expect chairs, editors, etc. come to you so much... by then it's often too late)

[dwsinger]

@marcoscaceres I think we have a problem here: my perception is that candidates who don't become visible either in their current role (e.g. as AC Rep, as a Chair presenting to the AC and driving discussions that involve more than their WG) or as a volunteer role (e.g. in PWE or ProcessCG) have a problem. It's not 'incumbent bias', it's 'visibility bias' – people are going to prefer candidates who have some track record, some visibility.

I don't think we can afford to appoint or elect people to the AB in the hope that they'll develop a track record there.

[michaelchampion]

One might argue that it would be better to have a AB selected from people who are active in working/interest groups than one selected from the people most widely visible to the AC (or most deeply supported by a sub-group in the AC) . That's how things work in the OpenJS Foundation https://github.com/openjs-foundation/cross-project-council (with which I'm affiliated at W3C). That would move W3C more toward being an explicit DO-ocracy, and reward people who do the most work (rather than those who do the most talking ;-) )

But alas, that's not the W3C we have. The best way I know to get elected to the AB is to make oneself visible to the AC as a reasonable, knowledgeable, concerned person who is willing to invest their time. But I have no clue how to convince one's management it's a good use of their budget (or family it's a good use of their personal time) to become electable, especially when someone could be writing/reviewing code or editing specs that others will write code to implement. 🤷🏽‍♂️

[frivoal]

This is tangential to the main point, but I think this needs a response, as I don't think it is good to leave undisputed a statement assigning blame in the wrong place.

As https://github.com/w3c/w3process/issues/589 illustrates, and this email shows, the AB has had blindspots about how changes to the Process impacts systems, tooling, and Editors directly

This is not due to a change in the Process, but in tooling, which the AB (or Process CG) are not in charge of. The tooling update was rolled out in in sync a Process change, but was not only (or even primarily) about the new Process. Yes, the system team chose to call that update "Process 2021", but that's up to them. I do not think it is fair to blame "AB blindspots" for the system team's software roll outs and insufficient communication with the rest of the ecosystem. The AB is not in charge of the system team, nor is it consulted about its roadmap and release schedule. I do no dispute that the AB has blindspots (who doesn't?) but that is not a relevant example.

the AB has left it to others to then deal with any fallout from changes to the Process document.

The AB does "leave it to others" in the sense that the AB is not the systems team, and does not run the W3C's infrastructure. But Process 2021 is fundamentally compatible with Process 2020. A few new things were possible, and a few corner cases were retired, but for the normal cases there was no necessity to break anything, and certainly not unannounced.

[npdoty]

The TAG has appointed seats, yes; one theory is that they exist to ensure that the TAG has the right mix of technical skills (security, back-end, user-facing, browser, etc.). I am not sure that the AB has the same issue. Note that we are introducing TAG appointed seat term limits.

I would suggest that the AB has similar reasons for an interest in a mix of skills and other qualities to ensure good advice and governance. Diversity of stakeholder groups, geography, demography, life experiences, industry sector experience, specific skills etc. would be helpful for the AB in its role of presenting advice to the W3C team about legal matters, process, conflict resolution and the interests of member organizations (not to mention potential members, or people affected by W3C activities).

That doesn't necessarily mean that term limits, quotas or appointments are the best or only methods to encourage that breadth of skills and representations. Elections by the membership can reflect that breadth, especially if the AC participates effectively and considers those in their rankings. But I do have some concern that election by W3C Membership may not alone be successfully meeting those needs (for the TAG, AB or BoD).

[mnot]

It's the right time to consider this. With all respect to the folks involved, we've had people serve more than a decade on the AB. Term limits are a widely-recognised, time-tested mechanism to assure that people don't get rusted into roles -- if we're going to rule them out, we need better reasoning than I'm seeing above.

[dwsinger]

I have a procedural concern with term limits (in general); in politics I observe that often people vote both for term limits and their long-standing incumbent, so I am led to suspect that they are trying to remove people that got elected, effectively, by others.

I have some anxiety about telling the community that they are not allowed to elect the person they perceive as best for the job, and also concern that promoting 'churn' is not good for us.

I would prefer to deal with the incumbent bias in other ways – notably:

• getting better at giving visibility opportunities to the membership,
• getting better at making it easier for the electing body to assess how effective an incumbent is

[michaelchampion]

I like @mnot 's characterization:

Term limits are a widely-recognised, time-tested mechanism to assure that people don't get rusted into roles

Asking incumbents to step away for at least one term every 8-10 years ensures there is some turnover, creates opportunities for new people to step up without having to compete against an insider, but allows the electorate to return an experienced person after a break IF they really are perceived as the best for the job.

@dwsinger wrote

promoting 'churn' is not good for us.

Agree in principle, since it takes a couple years to really figure out how to be effective on the AB, 2 or 4 year term limits would be counterproductive. But turnover every 8 or 10 years isn't "churn" as I understand it.

Also, term limits are a kindness to the people involved. I certainly needed a break and some fresh perspectives after 9 years on the AB! And it would be a good nudge for people to stay engaged in whatever career path brought them to the AB in the first place; few actual companies value "standards organization governance" as a career path any more.

[TzviyaSiegman]

if I'm understanding @michaelchampion correctly, he is suggesting limits on consecutive terms, not strict term limits. This is definitely an idea to float. We have conflicting needs to get new and fresh people serving and not lose the long view. I think starting with limiting consecutive terms is a good place to start.

[michaelchampion]

Yea I’m suggesting limiting to (4?) CONSECUTIVE terms. People could stay involved with “emeritus” status so their perspective not lost. And run in a future election if they wish .

But I also support @dwsinger ’s other proposals to limit incumbent bias.

[dwsinger]

I think Michael's proposal strikes a reasonable balance.

[cwilso]

I'd like to lead with "I have no personal interest in the outcome of this, because I will not run for the AB again after serving this new term."

But I do have thoughts, and I do bristle a bit at some of the characterizations here. As I did not crunch ALL the data from the entire history of our elections, forgive me if there are errors in my data, or omissions from early elections.

In a relatively brief scan, I believe there have been precisely three people who have successfully run in more than four consecutive elections:

• @chaals, who I believe ran successfully in six consecutive elections and served for eleven years from 2009-2020, winning a single-year seat in the last election he ran in. (He additionally is currently finishing a six-month partial term from running in a 7th election, after a two-year-plus break.)
• @michaelchampion, who ran for, won, and served five consecutive terms (for a total of nine years - one of them was a 1-year term) from 2010 to 2019.
• @dwsinger, who ran successfully for five terms in a row beginning in 2014 (resigning early in his fifth consecutive 2-year term to focus on his role on the Board of Directors).

(For overall service, Steve Zilles served as interim Chair for nearly 12 years, but (I think?) only ran in one election. As he was appointed chair, not an elected AB member, I've discounted him in this discussion; please don't take it as a discount of his service, however.)

In terms of who has RUN in consecutive elections, I believe @tantek and I are the winners - we both originally ran for the AB (and were elected) in 2013, and have run in every election we were eligible (eight, for me, and nine for Tantek) since then; however, we have both lost at least two elections during that time, skipping a year or more of service in the interim. Neither of us has served more than six years consecutively in our time on the AB (actually, neither of us has served more than five years consecutively yet; but I will have at the end of this term.) In terms of total service, at the end of the current term (presuming I serve the full term) I will have served on the AB for approximately ten and a half years (five terms plus a partial term), and Tantek will have served approximately nine and a half years.

Also of note are Judy Zhou and @LJWatson, who ran in and were elected in four consecutive elections, before resigning near the end of their fourth terms to serve on the Board of Directors.

The current members of the AB who have served the most consecutive terms are @frivoal and @TzviyaSiegman, who are half-way through their third consecutive terms. Avneesh Singh was just re-elected for a third term.

I believe in every election I've been around for, there has been at least one new member elected. For the two AB elections I lost as an incumbent AB member, I would point out that I was "replaced" by a new-to-the-AB member. (Natasha Rooney in 2017, and Tzviya Siegman in 2018.)

With that data - and there is a lot more, gleaned from the election archives and data on the AB page https://www.w3.org/2002/ab/ - I fail to see what real problem we have that we are trying to solve for by imposing term limits.

• Lack of "new blood" in each successive AB - provably not true; every election in recent memory has had a new member join.
• Single member serving for too long a consecutive time period - you are not fixing a problem that we are currently having; and in fact, there is only one person who has ended up SERVING more than nine years in a row, and only three who have served more than four terms in a row at all. "Asking incumbents to step away for at least one term every 8-10 years" is not a problem we have had; it would only have affected three people in our history.
• Single member RUNNING in more than four elections consecutively, whether they win or not - if your point is that members should not offer themselves to the Membership in service, well, frankly I think that's a bit insulting to those who offer.

With due respect, @mnot, no one currently serving on the AB would be affected by a four-consecutive-term limit, for at least three more years (if Tzviya or Florian re-ran successfully next year, after that term they would be unable to run again). Only one person in the history of the AB, as far as I can tell, has served as an elected AB member for more than a decade (Chaals), even in total time; only one more (me) will have done so at the end of the just-beginning terms.

I fail to see the need to impose more rules when there has been no need shown; I believe we have real problems to solve around elected positions (like onboarding, mentoring, getting more engagement out of those who are elected, and creating a pipeline for candidates people will participate in), and I think it would be far better to place our energies in trying to address those challenges than imposing term limits of any kind.

[mnot]

I can get on board with consecutive term limits, but 8 years is a long time.

The IETF's rule of thumb is that two terms is encouraged, three terms is to be looked askance at, and four terms is strongly discouraged in all but the most exceptional circumstances.

Because we have an election rather than a NOMCOM, it's not possible to apply a rule of thumb like that. My preference would be for limiting to two consecutive terms (i.e., take a break every third term); I could live with three.

I'll also point out that there are plenty of other things people can do in their "off" term now that we have eg a Board of Directors.

[dwsinger]

• @dwsinger, who served one and a half terms (1998-2001), then took a 13-year break, then ran successfully for five terms in a row beginning in 2014 (resigning early in his fifth consecutive 2-year term to focus on his role on the Board of Directors).

ah no. The David Singer who worked for IBM is another person (as is the David Singer who worked for Microsoft, but they didn’t do web work). So only the latter set of terms is me.

[cwilso]

Of course, my apologies @dwsinger (and corrected). You've mentioned that before, and the juxtaposition in the AB page always confuses me for some reason. You'd think I'd be aware that sometimes people have the same names. :)

[michaelchampion]

Thanks for the data @cwilso . They certainly discredit my suggestion for a 4-term limit, which would not solve many real problems! But to be fair, @marcoscaceres proposed a 5 year total limit, not an 8 year consecutive limit. 🤷‍♂️

Also, this issue is about "incumbent bias". Inspired by your data analysis, I crunched the 2009 - 2023 AB election results and found that incumbents who run have an 84% success rate, non-incumbents who run have a 45% success rate, almost a 2:1 advantage. That IS evidence of incumbent bias. And while I don't have hard data, I'm pretty sure there is a strong PERCEPTION of incumbent bias that discourages people from running.... "I'd like a term on the AB but the incumbents are all running again, so I'll lose".

The problems with incumbent bias noted in the original issue by @marcoscaceres include:

we just end up with the same ideas/people over and over again

OK, that's an exaggeration ... there is SOME new blood every election. But more "old blood" than new blood in most elections.(2014 and 2019 are the only AB elections since 2009 where more newbies than incumbents were elected).

limited chance for new people to get trained up in, or elected to, the AB.

I think this is the key issue here: How do we encourage people to engage to get "trained up" before running for the AB/TAG, and stay engaged in some sort of emeritus role so their knowledge isn't lost after they make way for those with fresh perspectives?

I'm not sure we need a process change, but I'd encourage a culture change along the lines of what @mnot mentions for IETF. After 2-3 terms, it's time for AB members to think hard about recruiting new people to run, mentoring them in a closely related group like the Process CG or or PWETF, stepping aside when they are ready to run, but staying engaged as with emeritus status as a source of hard-won wisdom and institutional memory.

[cwilso]

@michaelchampion it would be interesting if you could split the numbers by year, or at the very least show the data before 2017 and after (2017 election is when we switched to STV). If you were smarter than me and collected this data in a coherent form, let me know.

I would paint this as not "incumbent bias", but "incumbent bar". Ironically, this is easier to describe with terminology that would be more appropriate for approval elections, but: you're comparing success metrics for random candidates to success metrics for candidates who already garnered enough support to get elected once. That's hardly likely to ever be even close to even; the latter category has already passed the "approval bar" once before. The only bar for running as a candidate is some Member (the candidate themself, if they are a Member) is willing to nominate them. There are also other factors that will bump up some numbers.

At just a glance, I would point out what 2014 and 2019 had in common - LOTS of candidates, and few incumbents re-running. There were twelve candidates, for 5 seats with only 2 incumbents re-rerunning in 2014, and twelve candidates for seven seats with only 3 incumbents re-running in 2019. (In both these cases, it would have been impossible for more incumbents than new candidates to be elected.)

The culture change you suggest has already happened in a number of places - I've participated in several efforts to mentor and encourage new candidates, who have been elected, and if the TAG takes on their "expanded TAG reviewers" idea, that would be a great place for a "farm team". We've repeatedly encouraged AB candidates to do things like participate in the Process CG before; few candidates take us up on that. Heck, few candidates even reliably show up to Member meetings. Frankly, I'd be delighted if more people did, expressed an interest in maybe joining the AB, and asked how they could ramp up in participating in the work. A number of AB members - most DEFINITELY myself included - have offered assistance in this in the past; very few have taken us up on it.

Again, though, I believe imposing term limits is solving a non-problem - and although @mnot's suggestion might work well IF we had a nomination committee responsible for maintaining the health of the elected body (including mentoring), we do not have that; we have a very firm "proportional representation" voting system, with a sense that it will somehow result in the best wisdom. I fail to see how artificially ensuring that candidates serve for no more than, say, five years, would result in a more productive or useful AB (or TAG, or even Board for that matter.). But I would strongly caution that before any such change is made, it be clearly expressed exactly what we are trying to optimize for here.

[michaelchampion]

@cwilso the spreadsheet I'm using (hand-edited from reading AC-Members posts, could have errors, as could the formulas) is at https://docs.google.com/spreadsheets/d/19qZk2Ywt4oMs9wpv_SYwlHG0G1Sj9y4LLvYHHHYJI1w/edit#gid=0 and should be readable by all

The 2009-2017 percentages for newbies and incumbents winning percentage are pretty close to the 2009-2023 percentages, see the bottom couple of rows of the spreadsheet.

, I believe imposing term limits is solving a non-problem

I'm pretty neutral on term limits at this point. I think it is a problem to have slow turnover, and it would be good to have a stronger "bench" (in the sports sense https://www.linkedin.com/pulse/guide-common-sports-metaphors-laura-hale-brockway-els) of qualified people who could step into the AB and TAG, not to mention Board. But it would also be a problem if people are demotivated by feeling rejected or unfairly forced out. I strongly agree that the "farm team" (another US sports metaphor, sorry!) approach is a good one, but I am at least open to the possibility that formal term limits could give a nudge to incumbents to think about moving on.

But I would strongly caution that before any such change is made, it be clearly expressed exactly what we are trying to optimize for here.

Very strongly agree!

[frivoal]

found that incumbents who run have an 84% success rate, non-incumbents who run have a 45% success rate, almost a 2:1 advantage. That IS evidence of incumbent bias.

That's a 2:1 difference in outcome, not a 2:1 advantage. You could get that kind of result even if being an incumbent were to be a disadvantage (which I don't think is the case, but the number doesn't support the claim). For that to be a 2:1 advantage, you would have to assume that all of that difference is caused by being incumbent, which does not seem like a reasonable assumption.

Click here for excessive details playing with numbers

> When there are lots of candidates (and for some elections, there were, as @cwilso pointed out), then many candidates will fail, because there just aren't that many seats, and that can skews the statistics. There's a limit on how many incumbents can run, but not on how many new people can run. Taking an extreme example to show how the math works out, let's say that one year, there are 298 candidates (none of which are incumbents) for 4 seats. All the winners in that election are non-incumbents (but so are all losers). Then at the next election you have 6 people running for 4 seats, 4 of which are incumbents, 2 not. Let's say all the non incumbents win, and half the incumbents do. Averaged over these two years, the incuments have a 50% success rate (2 of 4), and non incumbents a 2% success rate (6 of 300). By your earlier wording, that as a 25:1 "advantage" for the incumbents, even though in that scenario, incumbents have lost _all_ the disputed seats. This theoretical exercise doesn't disprove an incumbent advantage, but the "2:1" claim doesn't prove it either. > > I've taken the data shared by @michaelchampion, and crunched it some more. https://docs.google.com/spreadsheets/d/1MgiN2awbWZQFsre-FCvRS4YGbWUj_pWmtvDwp_JfCAg/edit?usp=sharing. I've considered an incumbent seat "undisputed" if there aren't enough non-incumbents in a particular election to take it, and an non-incumbent seat to be undisputed if there aren't enough incumbents to take it. Disputed seats that those for which there are enough people in the election that it could go either way. With that, I find that 58% of disputed seats go to incumbents, and 42% to non-incumbents (or 56% vs 44% for the pre-2017 period). I'd say that's not too bad. Still, I'll agree that this is in excess of the proportion of incumbent vs non-incumbents going after these seats, but see the next point.

To be an incumbent, you need to be someone who (enough) voters like to vote for, or you wouldn't have been elected the previous time around. New candidates have not been subject to that filter yet. This does suggest that incumbents should be more likely to win than non incumbents, but that's selection bias at play, not necessarily because they are incumbents.

I don't think we have a problem of "no new blood": 6 of the 11 current ABers weren't on the AB 4 years ago.

I don't think we have a problem candidates consistently holding a seat for too long: based on https://www.w3.org/2002/ab/, the average tenure of people who're no longer on the AB is 4.7 years and and the median 3 years, or 4.6 and 3 if you include people who're currently on it (whose final term length is not necessarily known yet). And not all of that is consecutive.

If you look at gender statistic of the AB over recent years, we're not quite there yet, but the AB has been better balanced than the membership at large is

If you look at geographical balance of the AB, it is less concentrated than the membership at large is.

I'd say diversity in terms of company size, type of industry, disability, or age, also compare pretty well to the general membership.

This isn't to say that the AB is perfect, but these don't seem to be the characteristics of a body stuck in self perpetuation and in need of a forceful out.

I do think it is important that we work at supporting the diversity and renewal of our governance groups, but on that topic, more concerning to me would be:

• The low voter turn out in elections outside of a couple of geographies
• The very limited engagement in the Process-CG, the PWE-CG, or https://github.com/w3c/AB-memberonly/issues (which could serve as a "bench")
• The comparatively weaker diversity on the BoD
• The lack of diversity in upper management
• The lack of means invested by W3C to support diversity on the AB or TAG by sponsoring the travels of elected IEs (https://github.com/w3c/AB-memberonly/issues/1, https://github.com/w3c/board/discussions/42)

[dwsinger]

found that incumbents who run have an 84% success rate, non-incumbents who run have a 45% success rate, almost a 2:1 advantage. That IS evidence of incumbent bias.

That's a 2:1 difference in outcome, not a 2:1 advantage.

Strongly agree. It should not be a surprise that people who were electable before are more likely to be electable again. This is not evidence of bias at all.

[LJWatson]

I tend to agree with @cwilso in https://github.com/w3c/w3process/issues/697#issuecomment-1579648876.

Of the people on the AB when I joined in 2016, only three are still serving and, as @cwilso noted, none of them has served consecutive terms in the intervening years.

Of the people on the AB now (pre and post election), experience ranges from 0 to 6 consecutive years. Across 11 people, that seems like a reasonable balance.

From my own experience, it takes the first term to get familiar with the AB and how it works. It's only in the second term you feel able to really contribute and/or take the lead on one of its activities, and arguably a little longer before being ready to be nominated as co-Chair.

Building in an arbitrary circuit-breaker after two consecutive terms seems, well, like a solution in search of a problem to me.