User:Rw/Proposal:Welcome Working Group/2012-10-17 agenda/irclog
From OpenStreetMap Wiki
(02:00:33 PM) rweait: shall we *** begin logging *** ? (02:00:58 PM) rweait: Hello everybody. Say "hi" or something if you plan to participate (02:01:09 PM) pnorman: !ih (02:01:13 PM) mackerski: Evening (02:01:22 PM) RM87: hi (02:01:26 PM) SomeoneElse: Hello (02:02:07 PM) rieffel: hi (02:02:47 PM) rweait: Is there anything that anybody would like to add to the agenda for today? (02:03:15 PM) rweait: And, anybody who is new today, please feel free to introduce yourself. (02:03:17 PM) mackerski: rweait: Could you pop a link to that here, please? (02:03:24 PM) ***mackerski has just arrived hom (02:03:25 PM) mackerski: e (02:03:25 PM) rweait: http://wiki.openstreetmap.org/wiki/User:Rw/Proposal:Welcome_Working_Group/2012-10-17_agenda (02:03:29 PM) mackerski: Ta (02:03:46 PM) pnorman: Yes. I'm wondering about verifying the view that a large number of import meta-data tags discourage new editors from editing those objects. (02:04:04 PM) mackerski: Good topic (02:04:28 PM) alexbarth [~email@example.com] entered the room. (02:04:32 PM) rweait: Shall we add that to the end, then? (02:05:02 PM) mackerski: +1 (02:05:05 PM) rieffel: hey everyboy, i am rieffel, from-time-to-time mapper from germany, and by now trying to get more involved with osm, besides mapping. i read about this group on the ml and read as well the irc log of the last meeting, so the whole wg seems like a good catch for me to get more involved with osm (02:05:44 PM) rweait: Welcome. (02:06:36 PM) rweait: okay, Working group staffing. This group will need somebody to organize it in two ways. Someone to 'run' the meetings and someone to take care of minutes, reminders etc. (02:06:50 PM) rweait: And I think dividing it up helps with the load. (02:07:19 PM) rweait: We'll also want some software tools, at a minimum to track and graph our results. So, I think we'll want some coders committed to the group. (02:07:27 PM) rweait: And so now we need volunteers. (02:07:44 PM) rieffel: and to add my skills, i am not too bad in German, english and spanish/portuguese, do some python coding and am interested in contributing to the community (02:07:49 PM) ajashton [~firstname.lastname@example.org] entered the room. (02:08:37 PM) rweait: rieffel: would you be able to put an hour or two per week to the WG? (02:09:23 PM) rieffel: i think so. maybe i might need to skip a week or two, but generally speaking, yes indeed! (02:09:33 PM) rweait: Also, others feel free to join in. I'd like to see a volunteer and a back-p for both the chair and the secretary. (02:09:55 PM) rweait: rieffel: so I'll add you the the list of "coders" :-) (02:10:17 PM) rweait: I expect that we will all be interested in helping on the translation / cultural filtering aspects. (02:10:34 PM) rieffel: ok. and even more, i like to do translations and provide "cultural support" (02:10:38 PM) rieffel: ... (02:10:40 PM) mackerski: rweait: I have to plead ENOTIME for extra committed duties (02:10:40 PM) rieffel: :) (02:11:15 PM) rweait: mackerski: I already have you as board oversight and MT consultation, so I think you're well and committed. :-) (02:11:23 PM) mackerski: Indee (02:11:24 PM) mackerski: d (02:11:36 PM) rweait: I'm happy to take back up on either chair or secretary, but lead on neither. (02:11:55 PM) rweait: Too many other WGs (02:12:10 PM) rweait: Okay. Let's move on for now. We'll have to return to this today. (02:12:50 PM) rweait: Regarding our first experiment, "Early contact with new mappers is a benefit to mapper engagement" (02:13:15 PM) rweait: There is some prior art. What can we learn from it. Komzpa, can you share what you know? (02:13:53 PM) rweait: cquest, thank you for joining us, if you can tell us about your expereince as well? (02:14:02 PM) wonderchook [~email@example.com] entered the room. (02:14:18 PM) rweait: wonderchook: happy time zone. :-) (02:14:29 PM) cquest: I've been "welcoming" new contributors in France for at least one month (02:14:52 PM) wonderchook: Ha this week only rweait (02:14:56 PM) cquest: I wanted to see the effect of active contact thru private messages (02:15:48 PM) cquest: I'm also concerned about the ratio of people registering/doing 1 edit/doing a couple of edits/doing a lot of edits... (02:16:01 PM) cquest: Here's why I'm here tonight (02:16:16 PM) cquest: I'm also OpenStreetMap France association secretary (02:17:42 PM) rweait: What have you learned from the month of contacting new mappers? (02:18:06 PM) cquest: I got zero negative feedback, 20% positive, 80 no feedback (02:18:30 PM) rweait: positive feedback being what? (02:18:31 PM) cquest: I need to check again what happened to the people I contacted (02:18:36 PM) rweait: ah, okay. (02:18:51 PM) cquest: Positive... "thank you for contacting me, offering help, etc" (02:19:04 PM) rweait: Komzpa, please jump in when you can. (02:19:11 PM) rweait: cquest. excellent. (02:19:14 PM) cquest: Need to check if they did more edits tan thers... (02:19:27 PM) rweait: that was my next question. :-) (02:19:32 PM) cquest: I think the content of the first message they receive is very important. (02:19:49 PM) wonderchook: I would agree with that statement (02:19:50 PM) rieffel: cquest: would you mind giving me (apologies for my ignorace) a short example of this content? (02:20:02 PM) rweait: I agree. I think we see in the mailing list that a variety of message types can be successful. (02:20:05 PM) cquest: Mine was more an invitation to join the french community communication channels (mailing list, phpBB forum, etc) (02:20:32 PM) rweait: SomeoneElse: would you like to tell us about your contacts? (02:20:34 PM) rieffel: ok (02:21:09 PM) rweait: cquest, can you say how many accounts you contacted? (02:21:39 PM) SomeoneElse: I tend to message people within the English East Midlands if they're new and start making easily fixable mistakes (in hopefully a friendly apologetic way) (02:21:39 PM) cquest: Just a moment... I check my "outgoing box" (02:22:26 PM) SomeoneElse: Of roughly 90 new mappers in the last year in that area I sent roughly half a message (02:22:29 PM) pnorman: I've made some contacts in the past but don't have any way to get statistical significance out of them. My impression was it helped short-term but I'm not sure it showed any difference in mapper retention in the longer term (e.g. 6 months - 1 year) (02:23:06 PM) SomeoneElse: I haven't noticed a difference in "more likely to continue mapping" from messaged vs not messaged. (02:23:06 PM) rweait: pnorman: that's the sort of thing I hope that we can address by combining our efforts. (02:23:26 PM) RM87: When I have contacted newcomers, I usually sent some information about their first contributions, where to get help, where is the beginners guide and local community mailing list address. Results have been that less than 50% respond to this message. (02:23:30 PM) bfx [~firstname.lastname@example.org] entered the room. (02:23:52 PM) rieffel: same here. i made contact with some 5 new guys in my area where i spotted some errors. tried to be very freindly, avoiding condescending langauage and stuff. and almost all answered positively, aksed me for some more advice... (02:23:58 PM) rweait: SomeoneElse: you shared with me some anonymized results. I was surprised to find that 16 of 19 continued mapping. That seems very high / very successful. (02:23:58 PM) rieffel: *asked (02:24:02 PM) mackerski: I can't draw any conclusions from message sent either, but I've found that once you get new mappers really talking to more of the community (IRC is what works for this in Ireland), there is a very good hit rate (02:24:11 PM) mackerski: It's like they sense the momentum (02:24:39 PM) SomeoneElse: rweait: That's a selective group though - I tend to only message people after they keep making mistakes (e.g. P2 unjoined ways). (02:24:55 PM) SomeoneElse: It misses off the "let me just name my street" people. (02:25:00 PM) rweait: ah, so thaey are already not a "one and done" mapper. (02:25:34 PM) pnorman: Perhaps what is needed is a random study, finding new mappers in a selected area and then randomly selecting half of them for contact, then compairing the results after (02:25:54 PM) SomeoneElse: Feedback has been either neutral (i.e. just "ok") or positive. I don't think I've had any particularly negative feedback. (02:25:55 PM) AndroUser2 [~email@example.com] entered the room. (02:25:59 PM) rweait: I like that idea pnorman. (02:26:29 PM) wonderchook left the room (quit: Read error: Connection reset by peer). (02:26:59 PM) rweait: pnorman, how would you do it. (02:27:02 PM) rweait: ? (02:27:08 PM) AndroUser2 is now known as wonderchook (02:27:11 PM) rweait: email by hand... ? (02:27:19 PM) SomeoneElse: Also don't forget (in another context) that the licence email got quite a few former mappers going again for a while (02:27:40 PM) rweait: SomeoneElse: that might be a rich source of archival data to mine... (02:27:56 PM) cquest: crowdy outbox: around 1300 contacted (02:28:02 PM) pnorman: rweait: same as what cquest and others have been doing, just don't contact half of them and record their names instead (02:28:06 PM) rweait: cquest: wow. (02:28:24 PM) rieffel left the room (quit: Remote host closed the connection). (02:28:25 PM) SomeoneElse: My messages to new mappers are all personal and are about somewhere that I've got some knowledge of. (02:28:37 PM) cquest: We had a big peak in new contributors after some TV broadcast... (02:28:41 PM) rweait: If we wanted to track the results in an automated way, how would we do that? (02:28:46 PM) rieffel [~firstname.lastname@example.org] entered the room. (02:28:47 PM) pnorman: Basically, the same as now, but record non-contacts (02:29:22 PM) rweait: pnorman: can we collaborate and eliminate duplicate contacts, add randomization? (02:29:49 PM) wonderchook: Though we should probably randomize the contacts (02:30:23 PM) rweait: if we get help from sysdadmins, perhaps we can avoid knowing who were are contacting? (02:30:42 PM) rweait: It would be completely impersonal, but better contrlled. (02:30:45 PM) pnorman: wonderchook: the question is if contact from a local mapper improves retention (02:30:59 PM) rweait: pnorman, in part. (02:31:09 PM) rweait: also if contact in general is helpful. (02:31:39 PM) rweait: we may find, and SomeoneElse's results suggest that hyper-local contact is successful. (02:32:18 PM) mackerski: My licence change emails tend to support that (02:32:32 PM) mackerski: Pretty much all were lightly personalised to the region the user had been editing (02:32:33 PM) rweait: So do we tackle this by hand for a week or two and see what happens? (02:32:39 PM) pnorman: In the context of a medical study you need to randomize the contact because it's well known that regular contact with a doctor improves health. Here we don't know that - perhaps early contact scares people off (02:32:45 PM) rieffel: that is also what i thought. if you can actually refer to streets/pois etc they edited and they realize you are close by and not some sort of floating above higher instance (02:32:49 PM) AndroUser2 [~email@example.com] entered the room. (02:33:07 PM) rweait: rieffel: that has been my thought as well. (02:33:08 PM) SomeoneElse: http://pastebin.com/B8qnrqYi (02:33:10 PM) wonderchook left the room (quit: Read error: Connection reset by peer). (02:33:17 PM) AndroUser2 is now known as wonderchook1 (02:33:24 PM) rweait: Others on the lists have suggesetd that a contact that is too-personal, might seem creepy. (02:33:24 PM) SomeoneElse: (is my anonymised list of contacted or not) (02:34:01 PM) rweait: so we might test personal vs. form letter. vs. semi-localized., etc... (02:34:18 PM) rweait: but what to test first? and how to track it? (02:34:33 PM) rweait: If we're doing this individually, we'd like the results to be comparable. (02:34:35 PM) pnorman: The problem then is we need a much larger sample size to reach statistical significance (02:34:53 PM) cquest: I first sent personalized messages, but after the TV broadcast, could not do that anymore (300 newcomers on one day). Personalized ones were more successful (02:35:18 PM) RM87: we'll need an example letter to get this messaging going (02:35:21 PM) cquest: The result will also depend on local factors... (02:35:31 PM) rweait: cquest: your contacts were all across France, not local to your city? (02:35:44 PM) RM87: and some guides how much we can modify this letter (02:35:45 PM) cquest: all across France (02:35:49 PM) rieffel: Maybe we need to develop some kind of "standard" message that can simply be c/p?ed and just needs one or two modifications to sound very personal (02:36:08 PM) rieffel: ...in a good way (02:36:18 PM) rieffel: ...notcreepy way (02:36:26 PM) rweait: okay, we are running over scheduled time. Shall we continue anyway? (02:36:26 PM) mackerski: A Nigerian lad sent me such a mail only today (02:36:34 PM) SomeoneElse: http://pastebin.com/Gh7Bs2gy (02:36:43 PM) SomeoneElse: (my example message) (02:36:44 PM) rweait: mackerski: "saucy" (02:36:47 PM) mackerski: I'm good to keep going (02:37:38 PM) mackerski: SomeoneElse: Nice message (02:38:20 PM) SomeoneElse: That message gets a roughly 66% response I think (doing the sums in my head!) (02:38:35 PM) rweait: How about this? if you want to run an experiment this week, track the new users in your city / small region. contact half of them, but track all of them. (02:38:38 PM) AndroUser2 [~firstname.lastname@example.org] entered the room. (02:38:51 PM) AndroUser2 is now known as wonderchook2 (02:38:52 PM) rweait: Who would join in for that, this week? (02:39:17 PM) rieffel: this week, starting today, ending with the next meeting? (02:39:26 PM) rweait: Use one contact email form. only lightly-personalized. (02:39:26 PM) SomeoneElse: I can do that for a week, but numbers are unlikely to be huge (02:39:34 PM) rweait: rieffel: yes. (02:39:41 PM) cquest: I can spread that on the talk-fr mailing list (02:39:47 PM) rweait: what do we need to track? (02:39:57 PM) cquest: to get higher volume (02:40:01 PM) RM87: I'll probably get only 2 new users during that time, but I'm in (02:40:04 PM) rweait: cquest: perhaps not an army, until we know this is a positive? (02:40:04 PM) rieffel: rweait: maybe we should already all use the same message(probably localized) (02:40:49 PM) cquest: Ok, so... I'll ask for volunteers to help in this study (02:40:56 PM) cquest: a small army ;) (02:41:04 PM) SomeoneElse: Two other thoughts - (1) how to collaborate about who's looking at what area(s) (which fits in with the "social" stuff on the lists) (02:41:05 PM) rieffel: :) (02:41:09 PM) rweait: cquest: how will you eliminate duplicates? (02:41:23 PM) rweait: and overlaps? (02:41:29 PM) cquest: one volunter per area (02:41:33 PM) SomeoneElse: and (2) perhaps an impersonal "welcome! here's the help site, here's the wiki, here's IRC (02:41:35 PM) SomeoneElse: " (02:41:36 PM) SomeoneElse: ? (02:41:50 PM) rweait: SomeoneElse: sounds good. (02:41:57 PM) rieffel: just a stupid questions, since i did not work with users, but only worked with data in the past. is there a tool, which allows me to easily monitor my area for new users? (02:41:59 PM) cquest: tracking the effect of different kind of messages would be interesting too (02:42:09 PM) cquest: That's how ads are selected online (02:42:10 PM) wonderchook2 left the room (quit: Read error: Connection reset by peer). (02:42:12 PM) rweait: And if we commit to a single city / region now, we will avoid duplicates. (02:42:19 PM) wonderchook1 left the room (quit: Read error: Connection reset by peer). (02:43:01 PM) rweait: I'll take Toronto, Canada, if nobody else wants it. :-) (02:43:17 PM) SomeoneElse: rieffel: do a search for "neis-one" (02:43:20 PM) cquest: to gather statistical significant results, we need to replicate the test in different areas, and have one volunteer to deal with it in each area (02:43:43 PM) bfx: what about a list with people someone wrote to? (02:43:50 PM) rieffel: SomeoneElse: thanks. i knew in the back of my head that there was this tool, but couldnt come up with the name (02:44:08 PM) rweait: also #osm-bot on irc.oftc.net (02:44:33 PM) rweait: bfx: eventually, but for now, I think we need to keep out of each other's way. (02:44:46 PM) rweait: We'll need a tool to make this work on a larger scale. (02:44:56 PM) rweait: anybody else? (02:45:03 PM) rweait: cquest, your city? (02:45:04 PM) rieffel: i would take Duesseldorf and Dortmund in Germany (02:45:13 PM) cquest: I'm near Paris (02:45:18 PM) pnorman: http://resultmaps.neis-one.org/newestosm.php (02:45:41 PM) rieffel: thanks (02:45:54 PM) RM87: I'll take Estonia (02:46:09 PM) pnorman: With 13 in portland in the last week... it's like we got some publicity around there (02:46:16 PM) rweait: :-) (02:47:00 PM) bfx: i contact mappers east of rieffel (02:47:11 PM) rieffel: since the "new-mappers" map shows almost no activity for my area, i will also expand over the little villages where i grew up outside of Duesseldorf (02:47:34 PM) rweait: what should we track: username, user id, date of contact, type of contact email, response by email? (02:48:01 PM) rweait: And track which ones you did not contact as well. (02:48:11 PM) RM87: type of contact email? (02:48:17 PM) rieffel: maybe also a short description of the first edit? (02:48:25 PM) cquest: RM87: personalized or not, local or not, etc (02:48:33 PM) rweait: For now, that will be our uniform "Welcome, here is the help site" email. (02:48:43 PM) pnorman: I do not have the time to contact and there are not enough new mappers in western canada to be statistically significant - 1 in vancouver in the last week (02:49:01 PM) rweait: np pnorman. (02:49:56 PM) rweait: okay, still need volunteers for chair and secretary? (02:50:43 PM) rweait: Don't make me start "assigning" you. :-) (02:50:58 PM) pnorman: ENOTIME (02:51:00 PM) rweait: And pnorman, you wanted to talk about changeset tags? (02:51:11 PM) pnorman: I did? (02:51:17 PM) rweait: Didn't you? (02:51:24 PM) rweait: for bulk edits? (02:51:25 PM) pnorman: I wanted to talk about object tags (02:51:27 PM) rweait: okay. (02:52:13 PM) rweait: Let's presume that several of us will do the experiemnt and report back next week. (02:52:16 PM) pnorman: There is a perception that a large number of cryptic imported tags (e.g. TIGER) scares people from editing, but I'm not aware of any research to back this upe. (02:52:27 PM) rweait: Also, we'll presume same time / same channel next week. (02:52:42 PM) rweait: pnorman: can you suggest a test? (02:53:19 PM) pnorman: We might find that these tags discourage people from editing the objects, that they discourage people from editing the tags they don't understand or that they have no impact whatsoever (02:53:55 PM) cquest: pnorman: do these tags appear in Potlatch in simple mode ? (02:54:00 PM) rieffel: sorry, before we close this, maybe i missed it, but what about the message we are supposed to send to half of the new mappers? TO we think of our own? Will there be a template? (02:54:00 PM) RM87: As many don't see these tags in potlatch, these tags don't have an impact on absolute beginners (02:54:01 PM) SomeoneElse: I'm guessing newbies would mostly use P2 in simple mode and wouldn't see TIGER tags? (02:54:06 PM) rweait: historically, I can say that I've seen mappers faced with new TIGER data say, "I can't fix all of that" (02:54:36 PM) rweait: rieffel: shall we set that up by email tonight? (02:54:36 PM) pnorman: Oh yes, do we have any current stats on the first editor used? (02:54:51 PM) cquest: first editor ? Potlatch ! (02:54:53 PM) rweait: pnorman, can do. I don't have it now though. (02:55:03 PM) pnorman: cquest: what percentage? (02:55:13 PM) cquest: I would guess 99% ;) (02:55:15 PM) rweait: BAN FIRST EDITORS (02:55:41 PM) rweait: Might be intersting to put that on a time line as well. (02:55:49 PM) mackerski: pnorman: In Ireland most place nodes have rather a lot of importish tags (02:55:51 PM) cquest: That's one thing to keep track in our study as well as the kind of first edit (add a POI, changed something, etc) (02:55:56 PM) mackerski: I, um, may know the guy responsible for that (02:55:57 PM) rieffel: rweait: yes, we can do that (02:56:08 PM) pnorman: cquest: well, the problem with that is that until recently no one realized that PL1 was used by a significant percentage of first editors. I agree that I'd expect PL2 to be the most common, but numbers would be nice (02:56:25 PM) mackerski: I'm aware of many cases where new mappers just added another node with all the same stuff (02:56:35 PM) mackerski: Though that kind of thing happens a lot anyway (02:57:10 PM) SomeoneElse: pnorman: I can check the 90 or so (not all brand new) that I have details of (02:57:17 PM) rweait: editor usage graphs, by user. (one user at a time.) http://rweait.dev.openstreetmap.org/daily/rw__career.html (02:58:00 PM) pnorman: It's a fairly trivial postgres query against a changesetmd database to find out, I just haven't run it (02:58:41 PM) rweait: okay, I have changesetmd here. You too? (02:59:02 PM) cquest: editor on first edit(s)... (02:59:05 PM) rweait: So we're okay for next week, same time? (02:59:12 PM) cquest: that's what we're talking about, right ? (02:59:19 PM) rieffel: +1 (02:59:22 PM) mackerski: Should be OK (02:59:23 PM) rweait: cquest, yes, I think so. (02:59:31 PM) pnorman: cquest: ya (02:59:31 PM) RM87: ok (02:59:36 PM) cquest: +1 for next week (02:59:56 PM) rweait: okay. And let's draft our simple message by email tonight. rieffel: want to draft it? (03:00:00 PM) pnorman: I won't be making it next week - only here because I'm sick (03:00:19 PM) rweait: pnorman: conference cold? Sorry to hear that. (03:00:32 PM) lezurdis [~email@example.com] entered the room. (03:00:37 PM) rieffel: rweait: i could try. Just as a start, make it personal, but not creepy? (03:00:46 PM) rweait: yes. :-) (03:00:50 PM) rweait: not creepy. :-) (03:01:19 PM) rweait: "I see you are editing in $City" not "I was walking by your house..." (03:01:28 PM) rieffel: yeah... (03:01:58 PM) rweait: Any others planning to make contacts this week? Which cities? (03:02:10 PM) rieffel: but wouldnt we need two messages? one for a simple welcome mail and one for a simple welcome, but you did a little error email (03:02:34 PM) rweait: let's save that for another test. (03:02:40 PM) rieffel: bfx and me will cover a part of North-Rhine Westfalia, Germany (03:02:46 PM) rieffel: okay (03:02:54 PM) rieffel: start simple ad easy then (03:02:57 PM) rieffel: *an (03:03:00 PM) rieffel: d (03:03:07 PM) rweait: Okay, still no more volunteers? (03:03:15 PM) rweait: for chair and secretary? (03:04:09 PM) rweait: rieffel, what email address should we email you at for the draft? (03:04:34 PM) rieffel: use [email redacted] (03:04:43 PM) rweait: perfect. (03:04:48 PM) harry-wood [~Adium@18.104.22.168.dyn.plus.net] entered the room. (03:04:53 PM) rweait: okay, that's it everybody. see you next week.