The Emperor (Conflict Points) Has No Clothes

This morning I witnessed the immediate aftermath of a pretty bad accident on Woodall Rodgers Expressway (an ironic name since traffic rarely moves on it).  Nobody looked seriously hurt (thankfully) as far as I could tell, but the children looked pretty shook up.  However, one SUV was missing a front end. The other was missing the corner of its back end as several cars were also scattered all over the road with only one lane open for cars to get past.  Theoretically, there shouldn’t have been a collision here.  This is why the theory is wrong and we must design our roads and road systems with different (and better) priorities AND outcomes.

The dark red area is where the wreck occurred, even if the little crash icon is in the wrong place (unless there was another on 35)
The dark red area is where the wreck occurred, even if the little crash icon is in the wrong place (unless there was another on 35)

I didn’t see the actual wreck as it occurred so I can’t comment on what caused it.  Instead, I want to talk about how we end up with roads, like Woodall Rodgers where we try to funnel 150,000 to 200,000 vehicles through a confined space, which by design either happens at very high and dangerous speeds, or creates the congestion that causes this section of highway to be a very expensive parking lot for six hours a day.  Neither of these conditions is a good scenario.  So why then do we tolerate dangerous ineffective design for our very expensive infrastructure?  We have designed failure.  I’m going to theorize and analyze why.

Because of the 7 highways that converge at downtown Dallas, downtown functions (not as a downtown, but) as a giant interchange.  Few people are going to drive around the full perimeter of the inner highway loop (on purpose) unless they are lost or couldn’t make the right exit in time.

The reason is that these highways are a bypass, a way to cut through downtown.  They do this more than they deliver people TO downtown.  I see this to help describe the general movements past downtown, like so:


The nature of the interchange(s) and the width of the highway forces high speed lane changes
The nature of the interchange(s) and the width of the highway forces high speed lane changes


I show this because of a term we have heard recently and are likely to hear more of, “lane balance.”  Lane balance is in effect a weasel word to build more lanes.  In theory, the idea suggests that if four lanes from one highway meet four lanes of another highway, there should be then eight lanes to accommodate the converging 4+4 lanes and everything flows smoothly with no choke points.

Let’s ignore the issue of right-of-way width, acquisition, and the cost of land in downtown areas where this is likely to occur for the time being (even though that should kill the argument alone).

What the idea of lane balance and the consequence of converging too many highways in close proximity (say like right next to downtown, a place that if it ISN’T congested, is failing) is that they don’t take into account where the cars go next after they merge.  Even if you brought four lanes from one direction together and four lanes from another direction, the cars would all have to go in U-shapes to get to their destination.  Except, if you are driving north, you are unlikely to continue in a northerly direction.  If you are driving south, you are likely to continue south.

To describe this geographically so that you can picture it in your mind, if you are driving from Pleasant Grove past downtown, you are using Woodall Rodgers only if you are headed up Stemmons.  You wouldn’t use Woodall Rodgers to get to Oak Cliff, which you could do without having to make all of the lane changes that occur on Woodall Rodgers because there are better, more direct routes to get there.  In other words, if you are using Woodall Rodgers, chance are you have to make several lane changes.  This is why the road is so congested.  It is congested by design and because theory doesn’t match reality.

Drivers invariably have to criss-cross, which means lots of lane changes.  In the lane balance scenario, that means double the amount of lane changes.  For Woodall Rodgers, shown above, this means changing four lanes, in 0.8 miles, or roughly every ten seconds.  And this is where it gets interesting.

Our road network is a congested, over-wrought, unsafe mess because of the repetition of design for conflict points that ignores critical conflict points: lane changes.


This occurs at a local scale as well.
This occurs at a local scale as well.

Lane changes don’t rank highly on reasons for accidents, but that isn’t why they’re unsafe.  They also don’t rank highly in terms of severity of accidents when it is the cause because vehicles headed in the same direction and roughly similar speeds won’t have a lot of force in the direction of the collision (however, that doesn’t account for what happens afterwards, spinning out and such).

However, they are like the hockey assist of traffic accidents.  The pass before the pass that sets up the goal.  Patient zero of the malady.

The reason why lane changes, and in turn, not accounting for lane changes being dangerous and designing this danger out of the system, is because lane changes require you to take your eyes off the road (if you are being safe and driving defensively).  If, like the Woodall Rodgers example above, where you are forced to make a minimum of 4 lane changes in less than a minute, your eyes aren’t forward maybe at all.

Every time I’ve done this, I feel like I’m driving (fast) while only staring in the driver side rear view.  This is not safe, particularly as one interchange/choke point begins to back up into another and cars are stopped up ahead because they’re trying to exit onto Stemmons while I’m trying to get to the Hunt Hill Bridge.  And we’re all doing this ballet/demolition derby at 60 mph.


Conflict point diagrams are often used to demonstrate how roundabouts are better than conventional intersections, which sometimes, they actually are. But it depends on what your goals are.  Perhaps stopping cars occasionally is a good thing in some locations.  Therefore, judgment and establishing goals derived by area stakeholders is more important than theory.

Above is an example of a typical conflict point diagram.  Where the dots locate conflict points where paths collide.  You might say, “heck yeah, design to have as few as possible!”  Taken to its absurd logical conclusion, if we didn’t have the street at all, we would have zero conflict points and thus a perfect system.

Ignoring that hypothetical, we need to find the root of the problem in the real world.

These diagrams typically only show two lane roads with one lane in each direction.  If we showed four lane roads, like say Knox in its current configuration, it would look something like this:

Typical conflict point diagram for Knox as it currently exists
Typical conflict point diagram for Knox as it currently exists

I made this to show how all of the egress points to the various private parking lots and the valet stands located in the mid-block alley made for a number of unnecessary conflict points and thus less safe than if it were reconfigured as a three-lane road with a shared center turn lane, that actually isn’t shared because of the closed egresses.  The pedestrian conflicts are not removed (because pedestrians cross shopping is a good thing we want to encourage), but are instead lessened and controlled with mid-block cross and pedestrian refuge island.  Coincidentally, I further like to differentiate texture in designs where there are potential conflict points as a way to alert the driver to wake the f up and take their eyes off of their facebook update.

Knox after it goes on a diet.
Knox after it goes on a diet.


What I didn’t show in the top diagram, was conflict points from lane changes.  Why?  Because the diagram would be illegible.  I couldn’t make my point and there would be too many.  You also couldn’t locate any to make a dot because they could happen anywhere.  The conflict points created by multi-lane roads is thus INFINITE.

It means we can’t quantify these conflict points.  What do we do in a rational world where we can’t quantify something?  Everybody now, THAT’S RIGHT, we externalize it.  And what do we do with externalities kids?  THAT’S RIGHT!  We ignore externalities.

If you tried to map the conflict points of lane changes, the entire road would be a conflict point and by extension of the logic, again, we wouldn’t build the road.  And that would be irrational.  We can’t be irrational in a rational world.  Then you end up in the loony bin.

Since lane changes aren’t mapped and therefore not measured and accounted for, there is an impulse to make the road wider that it needs to be. Or at the very least, conflict points aren’t considered when the issue of congestion arises so the only impulse is to widen despite that we’re adding more infinite potential conflict points.  Then we end up trying to alleviate congestion by widening roads only to make more drivers and infinitely more conflict points on bigger, faster roads.  We have thus compounded the problem of our own creation.

Furthermore, we must ask the question what is really safer?  An intersection with many conflict points but what traffic there is, is moving very slowly?  Or, where conflict points are kept to a minimum but travel speeds are high?  Behavioral studies suggest that conflict points (in reality, not theoretically) are a measure of risk.  Where risk is high, we tend to behave more cautiously.  If you’re in an unfamiliar place and your mind isn’t turned onto auto-pilot (which is what roads are designed to do), you tend to slow down and be more aware of your surroundings.

When designing roads and road networks, there are always a number of important considerations (particularly in cities) that need to be balanced (and ideally benefited) such as economic development, congestion reduction, as well as impacts on modal choice and availability, route choice and availability, and trip length.

However, the number one consideration should always be safety.  And the measure of safety should be derived from in-depth and empirical methods rather than theory.

For example, we KNOW that network complexity/density as measured by intersection density increases safety.  Many intersections = many stops.  Many stops = slower traffic.  Slower traffic = less injuries and less severe ones.  Yes, collisions are a problem, and while we would all like none, we are a flawed species and are not very good, nor attentive drivers. I prefer non-injurious collisions to deadly maming ones.

20 is plenty. I know, I know. 20 mph won’t get you to your destination. But why, perchance, is your destination so far away?

But remember, intersections are by nature conflict points.  They are conflict points containing conflict points.  So if lots of conflict points are safer from a system-wide standpoint Should we even be using this measure at all?

Furthermore, one might think that if we prioritizing public safety, we might be sacrificing other considerations.  However, that isn’t true.  Highly interconnected grids mean more economic value and vitality as well as more capacity, since parallel routes are more efficient than the same lanes in one road.

In other words, a gridded system of narrow roads with few travel lanes is better in every meaningful way than what we’re currently designing and building when we prioritize all the wrong things.  Like this:

She's a real beaut'
She’s a real beaut’

Yes, people have died there.

Yes, people have died here in traffic collisions between 2004-13
Yes, people have died here in traffic collisions between 2004-13

And they tend to die more on wider roads despite there being theoretically no or at least fewer conflict points:

Too much color
Too much color

Because free flowing traffic is the be all and end all consideration (even if that free-flowing traffic occasionally collides into each other at high speed) we are left to wonder why Texas hasn’t gone a day without a traffic fatality in more than a decade.






  • Montemalone

    I wish the surface streets were maintained. Central Dallas already has the grid but it’s like driving on the moon.

  • kev4321

    How many people were killed in accidents while riding the bus? I think the author misses a very obvious alternative here. 1,000 buses could theoretically replace 40,000 cars at rush hour, per trip. There is no real solution for too many cars that doesn’t involve less cars.