A threshold guardian is often just a guy in the slow lane of the story appointing themselves traffic police. Like those truckers who decide to straddle merging lanes to enforce “fairness” when a full capacity zipper merge would in fact be more efficient.
Conversation
Replying to
I would like to see an example of what is meant by a zipper merge being more efficient.
1
Replying to
I understand (I think) how it'd be optimal efficiency if robots are driving/coordinating, but with humans I imagine the efficiency is lost to additional disruption in flow. Bandwidth through the choke point is maximized if flow is fastest.
1
I could probably just read the book but it's late and I'm tweeting from bed in the dark.
1
Replying to
The book references a paper. I don’t think human factor matters since it cancels out in both merge patterns, leaving net inefficiency of extra length of unoccupied lane causing backup to extend longer etc. And even if batch merges are better, blocking is still meaningless.
1
Replying to
Yeah this is why I'm curious to visualize an example, not just of the zipper but also of the less efficient alternative. I'm not sure what it's up against. (Agreed blocking lanes isn't productive.)
1
My assumption is that merges could happen earlier where traffic is less dense, which minimizes the impact on flow (speed) since humans are better then there's plenty of room. With an aim of maintaining maximum speed into and through the bottleneck, which determines bandwidth.
1
Found a paper referenced on Wikipedia, not sure if the same Tom references in the book. en.m.wikipedia.org/wiki/Merge_(tr
Suggests the efficiency gain is in (1) safety and (2) congestion disrupting less roadway because the traffic jam fills more lanes. That much is intuitive.
1
1
But also suggests that it doesn't improve bandwidth. And kind of handwaves the bandwidth problem as unrealistic to solve because humans go slow anyway 🤷
An assumption that, I think, makes the conclusion (that zipper is more efficient) more situational than general.
1
1
Replying to
Fine, ditch zipper. I’ll stand by claim that blocking harms the merge efficiency by shortening available coordination surface/line
Blocking is motivated by a fairness/cutting in line type silly instinct, there is no efficiency justification for even adding it to the solution space
1
Show replies

