[steindavidb]

It has to do a lot with reliability and cost. A smart car has a lot of expensive sensors and an expected life of a few years with regular maintenance. There's also a huge market opportunity around smart cars. Those cabinets can last decades, in all conditions, with minimal upkeep. Adding sensors decreases reliability and increases maintenance cost. There isn't usually an appetite to put traffic signal upgrades or extra maintenance headcount in city budgets for non-critical infrastructure upgrades, so the financial incentives to innovate are more limited as well.

[mojomark]

Thanks - interesting. I know traffic pattern/flow models and studies can be extremely complex. However, from the perspective of a 'dumb' commuter, it seems like the investment in smarter traffic signals would be offset by global savings from/economic benefits of reduced congestion (e.g. idle bus transit fuel/maintenance savings or generally increased standard of living and economic imprivements due to increased traffic flow).

[kqr]

It is useful to talk about both inefficiency (the thing you point out) and exploitability (the ability for any individual actor to make useful amounts of money out of fixing the inefficiency.)

Unless an area is both inefficient and exploitable, it is plausible that it will go unfixed for a long time.

When you start viewing it in those terms, you find obvious inefficiencies everywhere.