What more can a paid organizer do?
It may be that paid organisers simply increases the scale of the things they do already - eg. putting on more discussion groups, talks, workshops etc. though it could also be that having increased capacity enables groups to test promising strategies that they wouldn't have previously been able to.
One reason for thinking that it should be possible for organisers to increase the scale of their activities (and for this to result in an increase in the value that the group produces) is that even the largest groups seem to reach a fraction of their target audience. If groups aren't limited by the available target audience, and the grants process means that groups aren't limited by organiser time or funding, it seems that groups are likely to be able to increase the value they produce.
I'd be curious to know more about how people to message were selected
There weren't any strong guidelines in selecting people just encouraging people to talk to their friends. I chose people to message based on a combination of 1) how interested I thought they'd be (either based on previous conversations about EA or my knowledge of their interests) 2) how close we are, and I'd imagine others used similar heuristics.
and how the messages were crafted.
Here's a message I used that I also put up as an example for others, but there was an emphasis on making the messages personal rather than using a stock message, and so I expect that the type of messages that people sent varied quite a bit.
'Hey, last year i took the 'GWWC pledge' - a commitment to donate 10% of my income to the charities I believe are most effective at improving the world. I'd be really interested in hearing what you think about the idea and whether it's something you'd consider - what do you think? And do you fancy hearing a quick spiel about it? Anyway, what are you up to over New Year’s, when am I going to see you next?'
10% of messages converting to pledges is incredible, but potentially so incredible as to be suspicious.
The success that people have with this probably varies a lot. In particular having spoked to the person about effective altruism before made success a lot more likely. I think there was probably a fairly strong self selection effect, with those who have a lot of potentially interested friends being the people that decided to do the messaging and report their successes, and so I don't think the average GWWC member would be as successful (but probably still enough to make it worth doing).
Also the data from messaging friends seems consistent with the 1/25 message to pledge ratio from GWWC's previous attempts at messaging people - I'd expect messaging friends to be higher than this as the personal connection with the person you're talking about the pledge too seems to be quite an important factor.
Thanks for writing this up, lots of interesting ideas for retreat activities which I hadn't previously seen/ though of!