This site is an archive; learn more about 8 years of OpenHatch.

[Greenhouse] [lucas at debian.org: Surveying new package maintainers about their experience of contributing to Debian]

Asheesh Laroia asheesh at asheesh.org
Fri Jul 19 10:08:10 UTC 2013 

On Thu, 18 Jul 2013, Lucas Nussbaum wrote:

> Hi Asheesh,
>
> Given your general interest in increasing volunteer participation, it
> would be really great if you could review what I proposed on -project at .
> I can hold on sending the survey if you need more time. Just tell me.
>
> The mail below is the initial mail that provides the context, but
> I sent an updated proposal to -project@ later.

Hi Lucas! And a bunch of probably friendly other folks CC:d.

I seem to be having trouble falling asleep tonight, so you're in luck with 
an on-time reply to this after all.

I propose that before you send this, we take a little time (perhaps a week 
if we do it by email) to form a "Developer Advisory Team" (aka 
meta-mentorship team, aka any other name you like) within Debian and 
collaborate on some goals for it.

Some bits of preface text first.

Per the admittedly slightly vague 
https://lists.debian.org/debian-mentors/2013/07/msg00045.html , you should 
all know that David Lu (CC:d) is a GSoC student working with OpenHatch on 
some related tools. In particular, we're converting existing tools from 
Ubuntu that work from the ubuntu_upload_history in UDD (Ultimate Debian 
Database), and converting them into tools that work with Debian's similar 
data in upload_history. After that, we'll try to generalize the tools so 
they can work for not just Debian and package uploaders, but a variety of 
projects. Because are dreaming big and in a cross-project sense, I thought 
it made sense as an OpenHatch GSoC project. We currently intend to fold it 
into the openhatch.org code (which is free software) and help projects use 
it by running an instance on the web.

A year ago, I was inspired by the work of Ubuntu's semi-secret "developer 
advisory team. Their goals are:

'''This team in terms of UbuntuDevelopment, tries to fulfill the following 
tasks in the Ubuntu world:

* Reach out to new contributors, thank them for their work and get 
feedback.

* Reach out to people who might be ready to apply for upload rights and 
help them.

* Reach out to contributors that went inactive and get feedback from them 
and offer help.''' <https://wiki.ubuntu.com/DeveloperAdvisoryTeam>

The way they do the first one ("Reach out to new contributors...") is 
primarily through a tool that uses ubuntu_upload_history in UDD with the 
following workflow:

* Members of the "DAT" visit the website and see a list of people that 
have just done their first upload but who have not yet been greeted by the 
DAT

* They send an email in their personal mail clients to the new 
contributor, and they also annotate the web page in the "DAT web tool" so 
that it is clear to others that this person has been greeted

The way they track "...people who might be ready to apply for upload 
rights" is via the same workflow, but with a different query for people 
and the packages they have uploaded.

The way they track people who might have gone inctive is... yup same 
workflow, different query.

I also slightly exaggerate how secret the group is. I mean, it is clearly 
public; there is a wiki page on it, and they provide frequent reports. But 
they don't seem to go to conferences to publicly discuss how amazing their 
toolset is or write about it on the "metrics working group" list at 
http://www.theopensourceway.org/mailman/listinfo/metrics-wg .

One key point about Ubuntu DAT's app: because the tool is interactive, 
Ubuntu DAT members can consier the list of new contributors as a shared 
to-do list; if someone hasn't been pinged and greeted, then you know it is 
okay and useful for you to do so. By spreading the task of that initial 
pinging, it also means that the Ubuntu DAT strategy doesn't just create 
survey-type information; it creates lightweight mentor-type relationships. 
This is an advantage of the DAT way over the pure survey way, but actually 
DAT does also do a survey.

When I showed the earlier, Trello-based version of the DAT tool to Mako a 
few months ago, and he asked me, "Does it work?" Part of Dave's and my 
work this summer is to try to set up an experimental paradigm so we can 
find out the answer to that question. 
https://etherpad.mozilla.org/mentor-research documents some thoughts about 
that.

With all that said, what Dave and I have been doing is hacking the Ubuntu 
tool into something where Debian people can use similar workflow. We're 
part of the way there, and Dave can say more about how to join the beta 
test. That's what the "greenhouse" email list is for -- that beta testing.


I think that more important than our technical work this summer is the 
approach of Debian developers to these kinds of efforts. I think we should 
create a team around them, perhaps with similar structure to the "the QA 
team" in the sense that any DD is considered already a member, but perhaps 
with some explicit membership so that we can set some explicit, shared 
goals.

One key thing I have found in my surveying of other communities is that 
what new contributors say they want is not always the same as what causes 
change. So I really really encourage us to think first about behavioral 
studies, not about surveys.

The survey can be very useful, too. Indeed, Ubuntu's DAT's reports that 
they publish are based on a survey. I would like to see us get a random 
sample of new contributors in terms of who answers the survey questions. A 
good way to do that would be to randomly sample however we want, and then 
hound those people (gently) until they answer. This would give us a high 
response rate to a random sample, which provides more useful information 
than a response rate with unknown bias to a uniform sample. (For example 
-- if non-native English speakers really want a particular thing, but 
perhaps if they are the least likely to answer the survey, then doing the 
random sample plus hounding people to respond would get us their feedback. 
Simply sending it the survey as a broadcast would not get us their 
feedback.)

I also believe that membership in Debian's "Developer Advisory Team" (or 
whatever we wish to call it) would be a good stepping-stone to activity 
for DDs in the NM Front Desk. It is in some ways less burdensome to visit 
a web app and email people than to participate in NM, but it also would I 
think make you excited about NM. In that sense, perhaps Debian's DAT can 
be just be an offshoot of NM.

Having said all that, I will comment on your survey questions.

I liked your original survey more than the new one for how short it was. 
In particular, the new one is very long, and I suspect that will decrease 
your response rate. Also, visually, in my opinion, the checkboxes in an 
email make it burdensome on the eyes. But if you do try with a small 
sample like I said, and insist those people answer, that might be able to 
overcome that problem. So all in all, the new survey at 
https://lists.debian.org/debian-project/2013/07/msg00012.html is long but 
good.

We might send it to a small sample, just 5 at first, to get a sense of 
what the response rate is going to be like. If the response rate seems 
abysmal, we should alter the survey before sending it further.

I do think that the answers to these sorts of things are often best 
learned through interaction, not just surveys. When people have a contact 
point within Debian, they will answer these questions well. For that 
reason, you might ask that the most recent sponsor of these people's 
packages send them the survey.

The Ubuntu DAT strategy suggests that having even more contact points is 
of value. For that reason, there's a bug filed to help Debian DAT people 
send personalized emails asking for more info -- 
https://github.com/openhatch/oh-greenhouse/issues/9 is that bug.

I think that's all I can think of for now, before I fall asleep. I suppose 
I will join #debian-dat in irc.debian.org , but also you can reach me and 
Dave on #openhatch on irc.freenode.net. Unfortunately, for the next week, 
I am about to be swamped by going to the OSCON conference for a while, but 
I can make time to continue this conversation.

This email is open-ended for now; I'm curious what y'all think a good next 
step is. I can try to whip one up after the conference, though.

Oh, and I'm curious if we can offer Debian stickers or T-shirts to people 
to answer our survey.

Warm regards, and suggestions of strategy improvements very welcome,

-- Asheesh.

>
> Lucas
>
>
> ----- Forwarded message from Lucas Nussbaum <lucas at debian.org> -----
>
> From: Lucas Nussbaum <lucas at debian.org>
> To: debian-project at lists.debian.org
> Date: Wed, 17 Jul 2013 17:08:55 +0200
> Subject: Surveying new package maintainers about their experience of
> 	contributing to Debian
> Message-ID: <20130717150855.GA22577 at xanadu.blop.info>
>
> Hi,
>
> I feel that, as long-time contributors, we often lack a good view of how
> hard new people find it to get involved in Debian. In order to better
> identify blockers or difficulties that prospective contributors face
> when trying to contribute to Debian, I would like to survey new package
> maintainers, identified using [0] (thanks to Asheesh Laroia for fixing
> the UDD upload-history gatherer!).
>
> [0] http://udd.debian.org/cgi-bin/new-maintainers.cgi
>
> I started drafting the mail below, and would welcome reviews, comments,
> or additional ideas of questions. I plan to send the survey on 2013-07-19
> 12:00 (UTC).
>
> Lucas
>
> ------------------------------------------------------>8
> Subject: some questions about your experience becoming a Debian contributor
>
> Hi,
>
> You are receiving this mail because you recently had your first package
> accepted in Debian. First, congratulations, and thank you for contributing
> to Debian!
>
> In order to improve Debian's ability to welcome new contributors, I'd
> like to ask you a few questions about your own experience. Long-time
> members of the Debian community tend to forget how the process of
> getting involved in the project feels, and feedback from people who
> recently went through the process would be really useful.
>
> I will write and publish a summary of the collected answers. Your
> individual answers will not be made public, but might be shared with
> other DDs. Please be clear if you mention something that you prefer
> not to have shared too widely.
>
> So, here are the questions:
>
> Q1: Can you (briefly) introduce yourself? What motivated you to
>    start contributing to Debian? What are you (trying to) contribute
>    to in Debian?
>
> Q2: What are your reasons for starting to contribute to Debian *now*? Why
>    didn't you start before? :)
>
> Q3: What were the biggest surprises you had while going through the process?
>
> Q4: What did you find most difficult?
>
> Q5: Here are some ideas that have been mentioned, aimed at making it
>    easier for new people to start contributing to Debian. Could you
>    comment on them? If already implemented, would they have helped you?
> a) have clearer lists of easy bugs and tasks, suitable for new contributors,
>   so that people can learn about Debian by working on them
> b) have a "contributors' frontdesk" where people can describe their skills,
>   and get suggestions of areas of Debian where they can help
> c) develop internship programs (think of Google Summer of Code)
> d) develop peer-mentoring, where an experienced person guides a new
>   contributor
> e) do localized versions of debian-mentors (French, German, Spanish, ...),
>   where new contributors can ask questions about packaging
> f) organize IRC schools/seminars about packaging
>
> Q6: What other idea should be implemented, or what should be changed,
>    that would have made your life easier?
>
> Q7: Can I publicly list your name to thank you for participating in this
>    survey?
>
> Thanks!
>
> Lucas
>
> ----- End forwarded message -----
>
>

More information about the Greenhouse mailing list