One problem with swapcard is it is sequential. When I agree to a meeting with someone, we agree a time, even though there might be many slots that work for both of us. Later, I might want to meet with someone else, but the only slot that would work is the one I frivolously wasted earlier.
An alternative to this would be for people to set their general availability, declare bilateral meeting intentions (maybe with prioritization), and then for an algorithm to automatically assign people to timeslots and meeting locations in order to maximise connections and minimise overcrowding.
What should happen if I now want to also meet with someone else? Do my existing meetings get rescheduled? It seems to me that would cause people's timetables to be unpredictable, which will have annoying consequences like "The people from my uni group decided to have dinner together at 19:00 because that's when everyone happened to be free and now I can't make it", or any other thing that's not on the app.
You'd be required to immediately update any time you want to block because you have no manual control.
In the weeks leading up to the conference you enter the meetings you want to have, and block off your dinner date. Then at 1am on the morning of the conference the schedules are all set. You are correct that this does not work well if you want to schedule new meetings after the sorting hat has done its thing.
I don't know what the statistics are, but personally when I went to conferences most of my meetings were set after the conference had already started.
This piqued my curiosity, so I looked into it — it seems that there's a pretty equal split between pre-conference meetings scheduled (2449) and during-conference meetings scheduled (2039), with meeting confirmation frequency peaking on the Friday of the event. This is just EA Global: Boston 2023 data and I haven't looked at the others — but it does indicate to me that we don't want to set up systems that favour one type of user and disadvantaging another.
After having written an annual review of AI safety organisations for six years, I intend to stop this year. I'm sharing this in case someone else wanted to in my stead.
Hopefully it was helpful to people over the years. If you have any questions feel free to reach out.
Ahhh, what if you instead got substantial help, e.g., a research assistant? Mostly interested because there just aren't that many in-depth investigations of the whole field.
I think that the case for impact is now that OP/FTX read the review and change their perspective, rather than small donors. But I think that there is still value in this.
Yea; from my perspective, the total benefit of these posts might be higher than ever given the fact that there's much more activity and decisions being in the space now. That said, I imagine the costs are higher too (Larks's time is more valuable)
Thanks for pulling it together over so many years, and reading through a truly mindboggling array of papers & reports!