Every year, I help organize Minnebar, largest and longest-running technology unconference in…the US?
-
Every year, I help organize Minnebar, largest and longest-running technology unconference in…the US? North America? The Western Hemisphere? We’re not sure. It’s about 1000 people strong every year, and this is its 20th year. https://minnestar.org/minnebar/
The whole thing is volunteer-driven: people volunteer to lead sessions. There’s no application and no screening beyond the CoC; if you want to talk about it, raise your hand and you can.
This year, something interesting happened with the session topics.
🧵
-
Every year, I help organize Minnebar, largest and longest-running technology unconference in…the US? North America? The Western Hemisphere? We’re not sure. It’s about 1000 people strong every year, and this is its 20th year. https://minnestar.org/minnebar/
The whole thing is volunteer-driven: people volunteer to lead sessions. There’s no application and no screening beyond the CoC; if you want to talk about it, raise your hand and you can.
This year, something interesting happened with the session topics.
🧵
@inthehands Does that make you Minnbari?
-
Every year, I help organize Minnebar, largest and longest-running technology unconference in…the US? North America? The Western Hemisphere? We’re not sure. It’s about 1000 people strong every year, and this is its 20th year. https://minnestar.org/minnebar/
The whole thing is volunteer-driven: people volunteer to lead sessions. There’s no application and no screening beyond the CoC; if you want to talk about it, raise your hand and you can.
This year, something interesting happened with the session topics.
🧵
Here’s the list of sessions — remember, completely volunteer-driven, community-based, bottom-up:
https://sessions.minnestar.org/events/46/sessions
You’ll probably notice what everyone noticed: wow, that’s a lot of AI. Sheesh. It seems like half the thing is about AI! (It’s about 40%.) It sure seems like the whole conference is dominated by AI!
2/
-
Here’s the list of sessions — remember, completely volunteer-driven, community-based, bottom-up:
https://sessions.minnestar.org/events/46/sessions
You’ll probably notice what everyone noticed: wow, that’s a lot of AI. Sheesh. It seems like half the thing is about AI! (It’s about 40%.) It sure seems like the whole conference is dominated by AI!
2/
Concern about the flood of AI sessions was high enough that @iangreenleaf made this alternate session directory, with all the sessions hand-categorized as AI or not AI:
3/
-
Concern about the flood of AI sessions was high enough that @iangreenleaf made this alternate session directory, with all the sessions hand-categorized as AI or not AI:
3/
We don’t just have a lot of AI sessions this year. We have a lot of sessions, period. Typical years have 110, 130, 150 sessions. The previous record was 166. This year we had 193 (which melted down to 185 after a few people decided to cancel).
Take out the 40% of talks that are AI, and you have a Minnebar on the small side of normal. You might — with some serious squinting — say that this year we have a normal Minnebar •plus• a 40%-of-Minnebar-sized AI track added on. That’s not completely accurate, but it’s…accurate-ish.
4/
-
We don’t just have a lot of AI sessions this year. We have a lot of sessions, period. Typical years have 110, 130, 150 sessions. The previous record was 166. This year we had 193 (which melted down to 185 after a few people decided to cancel).
Take out the 40% of talks that are AI, and you have a Minnebar on the small side of normal. You might — with some serious squinting — say that this year we have a normal Minnebar •plus• a 40%-of-Minnebar-sized AI track added on. That’s not completely accurate, but it’s…accurate-ish.
4/
We’re not up to the interesting part yet.
My part in Minnebar is helping to create the schedule. We don’t hand-place talks in timeslots, or create “tracks” by topic. Instead, all the attendees vote, indicating •all• the sessions they’d like to see (as many or as few as they want to check off), and we generate a schedule from the votes. The schedule attempts to minimize attendee time conflict regret. That’s it.
(That process is interesting, but it’s not what this thread is about!)
5/
-
We’re not up to the interesting part yet.
My part in Minnebar is helping to create the schedule. We don’t hand-place talks in timeslots, or create “tracks” by topic. Instead, all the attendees vote, indicating •all• the sessions they’d like to see (as many or as few as they want to check off), and we generate a schedule from the votes. The schedule attempts to minimize attendee time conflict regret. That’s it.
(That process is interesting, but it’s not what this thread is about!)
5/
In short, this is what happens:
1. People volunteer to lead sessions
2. People vote for sessionsAfter step 1, it looks like AI was dominating. After step 2…not so much.
6/
-
In short, this is what happens:
1. People volunteer to lead sessions
2. People vote for sessionsAfter step 1, it looks like AI was dominating. After step 2…not so much.
6/
@inthehands oh good, that's what I was hoping you'd say next
-
Concern about the flood of AI sessions was high enough that @iangreenleaf made this alternate session directory, with all the sessions hand-categorized as AI or not AI:
3/
@inthehands @iangreenleaf Oh man, what an awesome (non-AI) lineup! I’ve gotta say “SOL LEWITT, COMBINATORIAL ENUMERATION, AND ROGUE” has me seriously considering looking at flights
-
Every year, I help organize Minnebar, largest and longest-running technology unconference in…the US? North America? The Western Hemisphere? We’re not sure. It’s about 1000 people strong every year, and this is its 20th year. https://minnestar.org/minnebar/
The whole thing is volunteer-driven: people volunteer to lead sessions. There’s no application and no screening beyond the CoC; if you want to talk about it, raise your hand and you can.
This year, something interesting happened with the session topics.
🧵
@inthehands You have childcare at your unconference?!
️ Best organizers ever. -
In short, this is what happens:
1. People volunteer to lead sessions
2. People vote for sessionsAfter step 1, it looks like AI was dominating. After step 2…not so much.
6/
There are approximately 1.5x as many non-AI sessions as AI sessions, but there were approximately ~2x as •votes• for non-AI sessions. The AI talks simply have not attracted as much interest.
This is especially striking when you look at the top sessions. AI sessions account for:
0 of the top 5 most-voted sessions
2 of the top 10
3 of the top 20
7 of the top 40(Accounting for which sessions were created earlier and thus collecting votes longer changes these results only slightly.)
7/
-
There are approximately 1.5x as many non-AI sessions as AI sessions, but there were approximately ~2x as •votes• for non-AI sessions. The AI talks simply have not attracted as much interest.
This is especially striking when you look at the top sessions. AI sessions account for:
0 of the top 5 most-voted sessions
2 of the top 10
3 of the top 20
7 of the top 40(Accounting for which sessions were created earlier and thus collecting votes longer changes these results only slightly.)
7/
To add insult to injury, one of those two talks about AI in the top 10 is titled “AI Sucks.”
Maybe Minnebar isn’t all about AI after all.
8/
-
@inthehands @iangreenleaf Oh man, what an awesome (non-AI) lineup! I’ve gotta say “SOL LEWITT, COMBINATORIAL ENUMERATION, AND ROGUE” has me seriously considering looking at flights
@grwster @iangreenleaf
You can ask @markgritter about it yourself! -
@inthehands You have childcare at your unconference?!
️ Best organizers ever.@mayintoronto
They are really great folks. -
There are approximately 1.5x as many non-AI sessions as AI sessions, but there were approximately ~2x as •votes• for non-AI sessions. The AI talks simply have not attracted as much interest.
This is especially striking when you look at the top sessions. AI sessions account for:
0 of the top 5 most-voted sessions
2 of the top 10
3 of the top 20
7 of the top 40(Accounting for which sessions were created earlier and thus collecting votes longer changes these results only slightly.)
7/
@inthehands I run a monthly product meetup with 5 concurrent sessions. Facilitators bring their own topics, and I curate the slate (and the facilitators).
I find that capping discussions of AI to 1-2 sessions means all the AI boosters go there, and the rest of the sessions can focus on their topics. It's bound to come up in every session, but it doesn't have to be the main topic.
-
To add insult to injury, one of those two talks about AI in the top 10 is titled “AI Sucks.”
Maybe Minnebar isn’t all about AI after all.
8/
There are a lot of possible explanations for this:
Maybe the people who aren’t interested in AI also happen to vote for more sessions. (At a quick prod at the data, penalizing votes from people who voted for large numbers of sessions does not seem to change the conclusions above, but I haven’t investigated this thoroughly.)
Maybe AI is a more fragmented topic due to its newness, and interest has not coalesced, leading to a large number of small sessions. (But again, •total• votes for AI, including the whole long tail, were proportionally smaller.)
9/
-
@inthehands @iangreenleaf Oh man, what an awesome (non-AI) lineup! I’ve gotta say “SOL LEWITT, COMBINATORIAL ENUMERATION, AND ROGUE” has me seriously considering looking at flights
@grwster hey, Greg! Thanks for the enthusiasm. You can see a shorter version of the talk at https://youtu.be/mtSjgxS-QS4 but I am still revising for Saturday. I'll share my new slides in GitHub.
-
There are a lot of possible explanations for this:
Maybe the people who aren’t interested in AI also happen to vote for more sessions. (At a quick prod at the data, penalizing votes from people who voted for large numbers of sessions does not seem to change the conclusions above, but I haven’t investigated this thoroughly.)
Maybe AI is a more fragmented topic due to its newness, and interest has not coalesced, leading to a large number of small sessions. (But again, •total• votes for AI, including the whole long tail, were proportionally smaller.)
9/
There are harsher interpretations too.
Maybe people are volunteering to give AI talks only because they feel like they’re supposed to. Maybe a lot of the AI talks are scam and spam. Maybe Those Darned AI People are all solipsists and just want to talk and don’t care whether anybody wants to hear it.
I’m hesitant to conclude anything nearly that judgemental from this data. There’s a lot of slippery questions about what exactly a “vote” means here, given how they’re collected and how people cast them.
10/
-
There are harsher interpretations too.
Maybe people are volunteering to give AI talks only because they feel like they’re supposed to. Maybe a lot of the AI talks are scam and spam. Maybe Those Darned AI People are all solipsists and just want to talk and don’t care whether anybody wants to hear it.
I’m hesitant to conclude anything nearly that judgemental from this data. There’s a lot of slippery questions about what exactly a “vote” means here, given how they’re collected and how people cast them.
10/
I will, however, draw two conclusions:
First, there’s an asymmetry between how many people will volunteer to •talk• about AI and how many people want to •hear• about AI. Without saying what it means, that asymmetry exists.
11/
-
There are approximately 1.5x as many non-AI sessions as AI sessions, but there were approximately ~2x as •votes• for non-AI sessions. The AI talks simply have not attracted as much interest.
This is especially striking when you look at the top sessions. AI sessions account for:
0 of the top 5 most-voted sessions
2 of the top 10
3 of the top 20
7 of the top 40(Accounting for which sessions were created earlier and thus collecting votes longer changes these results only slightly.)
7/
There are approximately 1.5x as many non-AI sessions as AI sessions, but there were approximately ~2x as •votes• for non-AI sessions.
Many talkers, few listeners. Curious.
