I’ve grumbled about the ludicrous price of conferences a lot recently.

[david_chisnall@infosec.exchange]

I’ve grumbled about the ludicrous price of conferences a lot recently. To put this in perspective:

I am off to a Dagstuhl seminar soon. The total cost of attending this three-day event (train to the airport, flight, taxi to the venue in the middle of nowhere, food, accommodation, and attendance) is less than just the registration fee for SOSP last year and EuroLLVM this year. And I expect to get far more out of it than either.

Last year I taught at a week-long summer school (PLISS). If I had paid the registration fee, the total cost of attending would have been slightly more than the cost of registration for either of the other events.

But both of those events are bigger! Yes, but does that make them more useful? I don’t have time to talk to most of the attendees and so much of the program is full of talks that could have been prerecorded videos (which I could watch without travelling) that the actual face-to-face time with other people is smaller.

Computer science really needs to rethink the ‘spend huge amounts of money going to massive conferences’ model. It excludes people who aren’t backed by well-funded institutions. And the value for attendees is quite low (far lower than for a lot of more focused smaller events).

#SOSP #EuroLLVM #ACM #Dagstuhl

[whitequark@social.treehouse.systems]

@david_chisnall also if you aren't there to meet people, then the carbon cost of it is basically indefensible (since watching a prerecorded talk is a strictly better option)

[pinskia@hachyderm.io]

@david_chisnall The GNU toolchain (GCC) cauldron is 1/6th of the price of EuroLLVM (and 1/10th the price of US LLVM dev conference) and held in the recent years (last 10 years or so) at an university [every other year in Prague].
Even with it being a smaller event it is still hard to meet everyone you wanted to talk with.
Plus there being held at an university has the benifit of meeting students interested in free software and compilers :).

[david_chisnall@infosec.exchange]

@pinskia I’ve suggested that we hold the LLVM dev meetings in a university where we can have the space for free, but the Foundation seems to be happy to just pay a lot. This made sense when most of the costs were paid by a corporate sponsor. The early US meeting were paid almost entirely by Apple, the Europe ones by Arm. You go through their event-organising systems get good rates, and they pay the cost, not the attendees. But they’ve kept the same model.

The last one where I looked at the costs was spending $30k on AV for a one-day event. I’ve organised conferences for 150 people with a total budget less than that.

I saw the invoice over someone’s shoulder for ASPLOS in Rhode Island a few years ago and the hotel was charging $50/head/meal for food. This was in downtown Providence, where it was completely surrounded by restaurants where you could eat much better food for $10-20. There was no need to cater lunch (most of the conferences I went to as a PhD student didn’t, unless they were too far from restaurants).

All of this is enabling naked profiteering by the venues, but it’s attendees not organisers who are paying.

[david_chisnall@infosec.exchange]

@whitequark

Yup. I haven’t been to a Dagstuhl seminar for a few years, but I had a lot more useful conversations at the last one than I did in any big conference I’ve been to because the entire event is structured to promote that.

[glipari@social.sciences.re]

@david_chisnall
I agree. IEEE Conferences are now around 1k euros, and the added value of participating is ridiculously low. Frankly, I am traveling much less now than I used to do (and that's good).