# Can a new shortcut slow everyone down? What does Berlin’s A100 show?

If enough drivers take the same shortcut, everyone can end up slower even when the number of trips stays the same. This is **Braess’s paradox**. Traffic on Berlin’s Elsenbrücke rose about 43% after the A100 extension opened. Whether travel times across the network suffered is open.[^1][^2]

[HTML version](https://onemorelane.org/en/faq/braess-paradox-berlin-a100/)

![Aerial view of crossing highways and curved connecting ramps.](https://onemorelane.org/photos/junction-960.webp)

Illustrative photo · [Emmanuel Appiah / Unsplash](https://unsplash.com/photos/Ry5z_fFKjFg)

## The example behind the paradox

The model starts with 4,000 people driving to the same destination at the same time. Route one begins with a narrow bridge and ends with a wide detour; route two begins with the detour and ends with a different narrow bridge. Each detour takes 45 minutes. Each bridge takes one minute for every 100 cars using it. If drivers split evenly, both routes take 20 + 45 = **65 minutes**.[^1]

Now a short one-way link connects the exit of the first bridge to the entrance of the second, and its own travel time is negligible. At first, taking both bridges saves time. Once everyone does it, each bridge takes 40 minutes, a total of **80 minutes**. Anyone switching back alone takes about 85 minutes. No individual gains by switching, even though everyone could gain together. Laboratory experiments with repeated route choices show the mechanism.[^1][^3]

## What Berlin counted on the Elsenbrücke

Berlin opened the 16th section of the A100 to Treptower Park on 27 August 2025. On the nearby Elsenbrücke, a twelve-hour count recorded 22,388 motor vehicles on 3 April 2025 and 32,031 on 28 April 2026. That is about **43% more**. Lane and weight restrictions applied on both dates.[^2]

The counter shows only how many vehicles crossed this one bridge. Proving the paradox would need travel times on the other routes under comparable demand. It is also possible that drivers lose time near Treptower Park while other destinations become quicker. Berlin’s response to parliament gives no representative before-and-after count across the wider network.[^2][^1]

## Test it in the simulator

Open a short second lane and watch what it does to the queue.

- [Open the short second lane](https://onemorelane.org/en/short-lane/)

## Related questions

- [Where does the extra traffic come from after a road is widened, and how can my trip still get quicker?](https://onemorelane.org/en/faq/do-more-highway-lanes-induce-demand/index.md)
- [Can a third lane help if the highway narrows back to two lanes?](https://onemorelane.org/en/faq/third-lane-before-a-two-lane-bottleneck/index.md)
- [Should an auxiliary lane end before or after the next highway entrance ramp?](https://onemorelane.org/en/faq/where-should-an-auxiliary-lane-end/index.md)

## Sources

[^1]: [Braess, Nagurney &amp; Wakolbinger (2005) · On a Paradox of Traffic Planning, route choice model](https://doi.org/10.1287/trsc.1050.0127)
[^2]: [Berlin Senate (2026) · One year after the A100 extension, parliamentary paper 19/26910, answers 1-2 (German)](https://pardok.parlament-berlin.de/starweb/adis/citat/VT/19/SchrAnfr/S19-26910.pdf#page=3)
[^3]: [Rapoport et al. (2009) · Two experimental tests of Braess’s paradox](https://doi.org/10.1016/j.geb.2008.02.007)
