The legality of changing lanes in an intersection may vary from state to state.
Crossing the street in the middle of the block is illegal in my state, and will cost you a large fine. You have to cross at a corner (and all traffic must stop for you or they can be ticketed, regardless of whether it’s a marked crosswalk or not. Every intersection is presumed to be a legal crosswalk). I know a little old lady, trying to catch her bus in the rain, who crossed the street in the middle of the block with no traffic around, and got ticketed for jaywalking (and missed her bus).
If you’re on a city street with four + lanes, and a vehicle going in your direction is stopped for what appears to be no reason, it may be because there’s a pedestrian/bicyclist crossing the road. Pay attention, slow a little, and be prepared to stop suddenly. Nearly killed someone when I learned this lesson.
In a long-past life, I had to (computer) model this exact situation (lane ends, when to merge) over a whole bunch of different variables, over several thousand runs. In slow traffic (defined as anything that drops you under the speed limit by 10% or more in the sim), total throughput through the system is basically unaffected by when people merge, although as an individual vehicle you can get through faster by merging later, at the expense of early mergers.
At speed, however, the simulations vastly support the “merge early” scenario: given sufficient early warning, you can achieve almost no slowdown of the traffic at all for anything by the heaviest of traffic: if you’re in the closing lane, there are likely to be a number of “ideal” merge opportunities during the marked interval, where you can move into a spot at no cost to speed on either side (this sim does require some variation in traffic speeds, but that’s almost always the case in the real world, and the variation can be tiny). The longer you wait to merge, the fewer these opportunities become, and eventually you won’t have any more. As soon as any vehicles hit this state, you get the aforementioned “shockwave” effect as people in the unclosing lane now have to explicitly allow the merge, which slows both lanes dramatically.
Interestingly, the one case that works the absolute best is the one we called “law abiding.” In most states, the law tells you that the responsibility of the driver in the unclosing lane is not to modify their driving to allow the merge, but rather to explicitly maintain their speed precisely to allow for maximum predictability by the mergers. In this scenario (drivers in the unclosing lane always maintain their speed and aren’t “helpful”), there is exactly zero slowdown for folks who are either in the unclosing lane or who merge early while opportunies exist. Throughput is maximal, and the entirety of the “penalty” is paid by the late mergers: if they run out of merge opportunities, they are forced to stop until a large opening occurs (perhaps indefinitely in the case of steady traffic, but that pretty much only happens in sims). So long as at least one merge opportunity exists per vehicle in the closing lane, early merging produces the “perfect” (well, maximal, anyway) result for both individual and total throughput in this scenario.
These results are the ones you’d obtain by straightforward logic (the simulations were to confirm a previously-calculated result), but they pretty much only work as long as everyone knows and follows the optimal scenarios. In the real world, results may vary.
TimeWinder, did you happen to model the effect of “open lane” drivers moving into the closing lane to stop the late mergers from passing the early mergers?
There’s been quite a bit of disagreement with this one. First of all, you don’t need a full car length of distance to maintain escape flexibility. And normally, drivers don’t pull right up behind the bumper of the car in from of them anyway. So we’re only talking about an extra yard or so for each car, which is probably immaterial to overall traffic in most situations.
You need to honk the Xanax-addled housewives driving their SUVs across two lanes out of their stupor, they will not avoid sideswiping your car just because they should be paying attention in a perfect world
If you are black, young, or from out of the county, your chance of getting a ticket increases exponentially
No. This was a long time (12+ years) ago, a graduate project used as an simulation algorithm design example for a class (that ended up only getting used for about ten minutes of discussion, despite the many, many hours of work. Such is computer science). I don’t recall all the variables, but we included at least: frequency of vehicles (both lanes), variability of speed and initial position (both lanes), several measures of “risk tolerance” (how much bigger than a car length did the space have to be for a car to be willing to merge, how close would a merged car have to be to make the car behind it in the merged lane slow down, etc.) and risk spread over the vehicle population, and variation of vehicle size (trucks and the like), over eight or ten different “merge algorithms”. Unless my memory fails me, all of these algorithms assumed that cars already in the unclosing lane stayed there, and had no effect on the closing lane cars except as obstacles, although we did model “helpfulness” - the cars on the unclosing lane slowing down (but never speeding up) to increase a gap space in front, statically and increasing in likelihood as the unclosing lane distance decreased. I don’t remember what effects each of these had, the results in my previous post are the only ones that stayed with me over time.
(And I need to reiterate; these were perfect, idealized computer sims, not real-world drivers: I post because I think the results were interesting, not because I believe they apply to the real world, which has a wide variety of inattentive, “greedy,” lacking in judgement, and simply malicious drivers in it. In particular, all sims have the inherent problem that all of their simulated entities are following the rules of the particular sim. In the real world, you never really know what others will do, and they’re very, very unlikely to be all following the same rules, anyway.)
Two other dot studies seem to disagree with Hampshires link
Bolding mine. A pretty high tech system, that reduced vehicle volume.
And
So it should be implemented when the traffic is at least 20% or more trucks or on 3 lane to 1 lane merges (if they can figure out a good system). Otherwise, not so good.
To me, a late zipper-merge seems obvious to be best-case scenario, but when you have everybody doing whatever they want, it messes everything up. Early-mergers annoy the living shit out of me, but I’m sure I annoy the shit out of people for merging at the merge point. If everybody just merged at the actual merge point instead of deciding where the merge point should be (and there’s usually several merge points, in my experience, when people early merge and that just fucks things up even more), everybody’s feelings can be spared and we’d have an orderly system.
Well, don’t know what to tell you. Early mergers are unhappy with you because you are slowing down traffic. If you would simply move over when the signs say that the lane is ending ahead, you would have plenty of places to merge and traffic could flow freely.
The problem is the early mergers merge something like a mile before the actual merge point (it’s more like there’s several merge points). Totally unnecessary and stupid. In my opinion, of course. And the way I view it, it’s not me slowing down traffic, it’s them. I’ve been lucky so far–haven’t noticed anyone trying to block me out or give me the finger. I used to be an early merger until I figured out nobody really seems to give a damn if I merge late and it makes absolutely no sense to merge early. If people don’t want me zipping ahead of them in a lane I’m perfectly entitled to (and, frankly, encouraged to use), they should stay in the damn lane until it runs out. Then everybody merges at the same time. Problem solved.
This is so obviously wrong (and I mean from a logical standpoint) that I wonder if we’re actually talking about the same thing. So, just to be clear:
The ideal is:
Traffic moves at the the maximal average speed.
No one has to slow down unnecessarily.
Throughput in terms of number of cars past the merge point/second is maximized.
Do we agree on this? Because given those constraints, and traffic that’s not so heavy it’s already slowed down, it’s easy to demonstrate that early merge is the right answer: by merging early (even if it’s a mile or so–in fact, especially if it’s a mile or so), the merger will have the maximum number of opportunities to merge into holes in the other lane without affecting traffic at all. Any car that does this has ZERO effect on the speed of traffic; they’ve just inserted themselves into an existing hole, and everyone keeps moving at the same speed. If everyone can do it, you have a perfect outcome (no one slowed down, everyone got merged).
The later you merge, the fewer of these holes exist, and eventually there won’t be any more at all. Now the last hole might happen right at the end, and the opportunity for a perfect “zipper” merge will allow zero-impact merging here, too. But assuming there are any cars at all in the lane to be merged into, there will not always be a hole available right at the end. And if there’s not, now a car has to slow down. Ideally it’s the merging car, but with real drivers on the road, you almost always get a car in the “open” lane that lets the late merger in. That triggers the worst case possible: the shockwave through the lane that shouldn’t have been affected at all.
And even if you get lucky and you got the hole right at the end of the ending lane, you’d STILL have been no worse off to have taken that hole earlier; there’s no expected payback to your gamble.
I’m talking about where there is a backup and traffic is moving slowly, not when traffic is flowing. In that case, a late zipper merge makes the most sense (as evidenced by most of the cites upthread.). Even Tom Vanderbilts best-seller on Traffic, entitled “Traffic, Why We Drive the Way We Do (And What It Says About Us)” comes to the same conclusion. When traffic is fairly free-flowing, then, yeah, I’ll merge earlier rather than later. If traffic is at a standstill, I’ll merge at the last possible point (in lane-ending situations, not at off ramp situations.)
Put it this way, I’m imagining a situation (the most common in my experience) where there are two lanes merging into one. They are both going slow enough that drivers have to make room for you and allow you in. Some drivers will be assholes and defend their space, some will cede space easily. Some mergers will merge early, some will merge late. You’ll have several different merge points (in my experience, there’s at least four or five places people merge.) I’m entitled to the left lane until the merge point, so why not use it? Now, if everybody did that, and we followed a zipper style merge (each car lettering one in), that’s the fairest solution for everyone involved. Nobody feels like their place in line was “jumped.” Everybody equally cedes one space to a vehicle when they get to the merge point, etc. When the traffic is built up, it doesn’t really matter where the merge point is, so much as that everybody merges in the same place and follows the courtesy of a zipper merge.
A good friend of mine grew up in California. He used to raise hell if I had my wheels turned while sitting at a red light. He claimed they give tickets for that in California. The concern is, if your hit from behind, your car will shoot straight through the intersection if the wheels are straight. If they are turned, you’ll go into the intersection crooked and cause a hazard.
I’d never heard of this in my driver training. But, he was really serious about it being a rule in California.
Almost certainly BS from your friend. A car will go where it’s going to go based on where it’s hit from. The direction that the wheels are turned barely comes into the factor and is not really worth worrying about. Same with the puting it in neutral and the parking brake at a stop light. The lack of car and physics knowlege in this thread has been quite astonishing,