Epilogue: From Consistency to Evidence

Consistent application is the missing link in curve advisory speed practice. SafeCurves is an automated and computational platform for applying curve advisory speed standards at network scale, and that matters because it delivers consistency. It cannot, however, create uniformity on its own, because it does not determine the governing technical standard. That remains the responsibility of the jurisdiction, whether it adopts the American Association of State Highway and Transportation Officials (AASHTO) standard, the MUTCD standard, or some other technical standard. In fact, anecdotal practice suggests that variation can persist even within a single state, where different districts or regions may apply different thresholds or interpretations of the standard – it is a minefield of varying practice.

SafeCurves removes much of the variability that has traditionally undermined curve advisory speed setting and allows the same technical logic to be applied systematically across an entire network. That alone is a material step forward.

But consistency is not the end goal - it is the enabling condition.

The deeper issue remains unresolved. While agencies have invested decades refining advisory speed procedures, we still lack a clear, evidence-based answer to a fundamental question: which advisory speed standard produces the best safety outcome?

SafeCurves creates a genuine opportunity to answer that question.

Because it can apply different technical standards consistently and at scale, it allows for meaningful comparison between them - something that has historically been difficult, if not impossible, using field-based methods. For the first time, we can isolate the effect of the standard itself, rather than the variability in how it is applied.

This opens the door to a more rigorous form of analysis:

• Comparing cohorts of jurisdictions using different advisory speed standards

• Controlling for key variables such as geometry, operating speed, and out-of-context curves

• Testing not just compliance, but actual safety outcomes, including fatal and serious injury risk

Importantly, this reframes the problem. The question is no longer whether an individual advisory speed is “correct” in isolation. The question becomes whether a standard - applied consistently - delivers a measurable safety benefit at network scale.

That is the shift from craft to evidence.

There are credible hypotheses to test. A more conservative standard may improve safety by identifying genuinely hazardous curves, even if compliance is lower. Conversely, a less conservative standard may improve credibility and compliance, but risk missing critical locations. Over-signing may dilute driver attention, while under-signing may fail to alert drivers when it matters most. Each of these outcomes is plausible. None has yet been resolved conclusively.

SafeCurves provides the analytical platform to test them properly.

However, doing so will require intent. It will require collaboration across jurisdictions, agreement on comparable datasets, and a willingness to evaluate long-standing practice against measurable outcomes rather than tradition or convenience. It will also require normalising for the factors that most strongly influence curve safety - particularly the prevalence of out-of-context curves - so that conclusions reflect the effectiveness of the standard, not the characteristics of the network.

The opportunity is significant.

If we can use consistent computational methods to evaluate advisory speed standards against real-world safety performance, it should be possible to converge on a single, evidence-based approach - one that restores the uniformity that frameworks like the Manual on Uniform Traffic Control Devices (MUTCD) have always sought, but never fully achieved.

That would represent a step change. Not just better signing, but a better understanding of what works.

So, the question is no longer whether consistency can be improved. It can.

The question is whether we are willing to use that consistency to test our assumptions - and, if necessary, change them.

Because if advisory speeds are a critical part of how drivers read the road, then getting them right is not just a technical exercise. It is a safety outcome. And we now have the tools to test that properly.