# The Proof of Innocence

> An annotated edition of this paper on Fermat's Library, where scientists and readers explain its ideas, equations and history in the margins. Annotated edition: https://fermatslibrary.com/s/the-proof-of-innocence

- **Authors:** Dmitri Krioukov
- **Published:** 2012
- **Topic:** Mathematics
- **Original:** https://s3-us-west-1.amazonaws.com/fermatslibrary/uploads%2F1696171619145-3xqnsba78i7-e23309f3008e20275a597e05dcf71931%2F1204.0162v1.pdf
- **Pages:** 4
- **Annotations:** 6 comments in 6 threads
- **Featured:** Fermat's Library Journal Club
- **Annotated edition:** https://fermatslibrary.com/s/the-proof-of-innocence

## Annotations

### Annotation 1 (page 1)

Beside the passage: “Dmitri Krioukov 1 1 Cooperative Association for Internet Data Analysis (CAIDA), University of California, San Diego (UCSD), La Jolla, CA 92093, USA”

**Annotation** by Luís Batalha (2023-10-01):

> ### TL;DR
>
> In 2012, Dmitri Krioukov, a physicist at the University of California, San Diego, faced a $\$400$ fine for allegedly running a stop sign. Instead of accepting the fine, Krioukov crafted an academic paper titled "The Proof of Innocence" to argue his case in court.
>
> Surprisingly, the judge was convinced and acquitted him. Krioukov shared his paper online with a subtitle, "A way to fight your traffic tickets," humorously noting that the paper earned a special prize of $400, which he didn't have to pay to California.
>
> The paper discusses a scenario in which if a car stops at a stop sign, an observer (e.g., a police officer) located at a specific distance perpendicular to the car's trajectory might perceive the car as not stopping under specific conditions. These conditions are:
>
> 1 - The observer measures not the linear but angular speed of the car.
>
> 2 - The car decelerates and subsequently accelerates relatively quickly.
>
> 3 - There is a short-time obstruction of the observer's view of the car by an external object (e.g., another car) at the moment when both cars are near the stop sign.
>
> Disclaimer: the paper was submitted on April 1st, 2012, so there's a non-zero probability this could a be joke by Dmitri :) Either way, it's a good paper.

### Annotation 2 (page 1)

Beside the passage: “S (stop sign)C (car)”

**Annotation** by Luís Batalha (2023-10-01):

> Prof. Dmitri Krioukov graduated with a Diploma in Physics from Saint Petersburg State University in 1993, and earned his Ph.D. in Physics from Old Dominion University in 1998. He transitioned to the industry and later returned to academia, joining CAIDA at UCSD as a Senior Research Scientist. Since 2014, he's been an Associate Professor at Northeastern University, engaging in complex network theory research as the Director of the DK-Lab at the Network Science Institute.
>
> ![](https://coe.northeastern.edu/wp-content/uploads/profiles/affiliated/krioukov-d.jpg)

### Annotation 3 (page 1)

Beside the passage: “v x”

**Annotation** by Omar Rabie (2024-05-09):

> He assumed the police officer measures the angular velocity which is rarely happening

### Annotation 4 (page 2)

Beside the passage: “0.1”

**Annotation** by Luís Batalha (2023-10-01):

> [Here's an applet to illustrate dynamically the two situations that the author describes](https://www.geogebra.org/m/uh8kqbf4). You can alter the acceleration of the car and the distance away the
> observer is to see (from the arrows) how the angular speed changes.

### Annotation 5 (page 3)

Beside the passage: “Finally, we consider what happens if the O’s observations are brieﬂy obstructed by an external object, i.e., another car, see Fig. 4 for the diagram depicting the considered situation. The author/defendant (D.K.) was driving Toyota Yaris (car C 1 in the diagram), which is one of the shortest cars avaialable on the …”

**Annotation** by Luís Batalha (2023-10-01):

> Krioukov told PhysicsCentral that the case and his argument were real, but that he left a flaw in his work for others to find.
>
> According to the manufacturer’s data, Toyota Yaris accelerates from 0-100 km/h in 4.8 seconds = 5.8 $m^2s^{-1}$, making it hard to believe in the 10 $m^2s^{-1}$ suggested by the author.

### Annotation 6 (page 4)

Beside the passage: “3. the O’s view of C 1 was brieﬂy obstructed by another car C 2 around time t = 0.”

**Annotation** by Luís Batalha (2023-10-01):

> The conclusions make sense, but only if you accept that the length of the car that obstructed the view of the police officer was similar to that of a Subaru Outback :)
> ![](https://cdn11.bigcommerce.com/s-pusehjx/images/stencil/1280x1280/products/6626/21101/ironman-Outback-woods-profile_1500x1000__93093.1683670385.jpg?c=2)

## Full text

Extracted from the PDF. [Annotation N] marks where each annotation above sits.

#### Page 1

The Proof of Innocence

Dmitri Krioukov [Annotation 1] 1 1 Cooperative Association for Internet Data Analysis (CAIDA), University of California, San Diego (UCSD), La Jolla, CA 92093, USA

We show that if a car stops at a stop sign, an observer, e.g., a police oﬃcer, located at a certain distance perpendicular to the car trajectory, must have an illusion that the car does not stop, if the following three conditions are satisﬁed: (1) the observer measures not the linear but angular speed of the car; (2) the car decelerates and subsequently accelerates relatively fast; and (3) there is a short-time obstruction of the observer’s view of the car by an external object, e.g., another car, at the moment when both cars are near the stop sign.

I. INTRODUCTION

It is widely known that an observer measuring the speed of an object passing by, measures not its actual linear velocity by the angular one. For example, if we stay not far away from a railroad, watching a train approaching us from far away at a constant speed, we ﬁrst perceive the train not moving at all, when it is really far, but when the train comes closer, it appears to us moving faster and faster, and when it actually passes us, its visual speed is maximized.

This observation is the ﬁrst building block of our proof of innocence. To make this proof rigorous, we ﬁrst consider the relationship between the linear and angular speeds of an object in the toy example where the object moves at a constant linear speed. We then proceed to analyzing a picture reﬂecting what really happened in the considered case, that is, the case where the linear speed of an object is not constant, but what is constant instead is the deceleration and subsequent acceleration of the object coming to a complete stop at a point located closest to the observer on the object’s linear trajectory. Finally, in the last section, we consider what happens if at that critical moment the observer’s view is brieﬂy obstructed by another external object.

II. CONSTANT LINEAR SPEED

Consider Fig. 1 schematically showing the geometry of the considered case, and assume for a moment that C’s linear velocity is constant in time t,

v(t) ≡ v 0 . (1)

Without loss of generality we can choose time units t such that t = 0 corresponds to the moment when C is at S. Then distance x is simply

x(t) = v 0 t. (2)

Observer O visually measures not the linear speed of C but its angular speed given by the ﬁrst derivative of angle α with respect to time t,

ω(t) = dα dt . (3)

O (police officer)

S (stop sign)C (car) [Annotation 2]

r 0

L (lane)

α

v [Annotation 3] x

FIG. 1: The diagram showing schematically the geometry of the considered case. Car C moves along line L. Its current linear speed is v, and the current distance from stop sign S is x, |CS| = x. Another road connects to L perpendicularly at S. Police oﬃcer O is located on that road at distance r 0 from the intersection, |OS| = r 0 . The angle between OC and OS is α.

To express α(t) in terms of r 0 and x(t) we observe from triangle OCS that

tan α(t) = x(t) r 0

, (4)

leading to

α(t) = arctan x(t) r 0

. (5)

Substituting the last expression into Eq. (3) and using the standard diﬀerentiation rules there, i.e., speciﬁcally the fact that

d dt arctan f(t) = 1 1 + f 2

df dt , (6)

where f(t) is any function of t, but it is f(t) = v 0 t/r 0 here, we ﬁnd that the angular speed of C that O observes

arXiv:1204.0162v1 [physics.pop-ph] 1 Apr 2012

#### Page 2

2

−10 −5 0 5 1 0 0

0.2

0.4

0.6

0.8

1

t (seconds)

ω(t) (radians per second)

FIG. 2: The angular velocity ω of C observed by O as a function of time t if C moves at constant linear speed v 0 . The data is shown for v 0 = 10 m/s = 22.36 mph and r 0 = 10 m = 32.81 ft.

as a function of time t is

ω(t) = v 0 /r 0

1 +



v 0 r 0

 2

t 2

. (7)

This function is shown in Fig. 2. It conﬁrms and quantiﬁes the observation discussed in the previous section, that at O, the visual angular speed of C moving at a constant linear speed is not constant. It is the higher, the closer C to O, and it goes over a sharp maximum at t = 0 when C is at the closest point S to O on its linear trajectory L.

III. CONSTANT LINEAR DECELERATION

AND ACCELERATION

In this section we consider the situation closely mimicking what actually happened in the considered case. Speciﬁcally, C, instead of moving at constant linear speed v 0 , ﬁrst decelerates at constant deceleration a 0 , then comes to a complete stop at S, and ﬁnally accelerates with the same constant acceleration a 0 . In this case, distance x(t) is no longer given by Eq. (2). It is instead

x(t) = 1 2 a 0 t 2 . (8)

If this expression does not look familiar, it can be easily derived. Indeed, with constant deceleration/acceleration, the velocity is

v(t) = a 0 t, (9)

−10 −5 0 5 1 0 0

0.1 [Annotation 4]

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

t (seconds)

ω(t) (radians per second)

a 0 = 1 m/s 2

a 0 = 3 m/s 2

a 0 = 10 m/s 2

FIG. 3: The angular velocity ω of C observed by O as a function of time t if C moves with constant linear deceleration a 0 , comes to a complete stop at S at time t = 0, and then moves with the same constant linear acceleration a 0 . The data are shown for r 0 = 10 m.

but by the deﬁnition of velocity,

v(t) = dx dt , (10)

so that

dx = v(t) dt. (11)

Integrating this equation we obtain

x(t) =

Z x

0

dx =

Z t

0

v(t) dt = a 0

Z t

0

t dt = 1 2 a 0 t 2 . (12)

Substituting the last expression into Eq. (5) and then differentiating according to Eq. (3) using the rule in Eq. (6) with f (t) = a 0 t 2 /(2r 0 ), we obtain the angular velocity of C that O observes

ω(t) =

a 0 r 0 t

1 + 1 4



a 0 r 0

 2

t 4

. (13)

This function is shown in Fig. 3 for diﬀerent values of a 0 . In contrast to Fig. 2, we observe that the angular velocity of C drops to zero at t = 0, which is expected because C comes to a complete stop at S at this time. However, we also observe that the higher the deceleration/acceleration a 0 , the more similar the curves in Fig. 3 become to the curve in Fig. 2. In fact, the blue curve in Fig. 3 is quite similar to the one in Fig. 2, except the narrow region between the two peaks in Fig. 3, where the angular velocity quickly drops down to zero, and then quickly rises up again to the second maximum.

#### Page 3

3

O (police officer)

S (stop sign)C 1 (car #1)

r 0

L 1 (lane #1)

L 2 (lane #2) C 2 (car #2)

l 1

C 1 = Toyota Yaris

l 1 = 150 in

l 2

C 2 ≈ Subaru Outback

l 2 = 189 in

l 2 − l 1 = 39 in ≈ 1 m

FIG. 4: The diagram showing schematically the brief obstruction of view that happened in the considered case. The O’s observations of car C 1 moving in lane L 1 are brieﬂy obstructed by another car C 2 moving in lane L 2 when both cars are near stop sign S. The region shaded by the grey color is the area of poor visibility for O.

IV. BRIEF OBSTRUCTION OF VIEW AROUND

t = 0.

Finally, we consider what happens if the O’s observations are brieﬂy obstructed by an external object, i.e., another car, see Fig. 4 for the diagram depicting the considered situation. The author/defendant (D.K.) was driving Toyota Yaris (car C 1 in the diagram), which is one of the shortest cars avaialable on the market. Its lengths is l 1 = 150 in. (Perhaps only the Smart Cars are shorter?) The exact model of the other car (C 2 ) is unknown, but it was similar in length to Subaru Outback, whose exact length is l 2 = 189 in. To estimate times t p and t f at which the partial and, respectively, full obstructions of view of C 1 by C 2 began and ended, we must use Eq. (8) substituting there x p = l 2 + l 1 = 8.16 m, and x f = l 2 − l 1 = 0.99 m, respectively. To use Eq. (8) we have to know C 1 ’s deceleration/acceleration a 0 . Unfortunately, it is diﬃcult to measure deceleration or acceleration without special tools, but we can roughly estimate it as follows. D.K. was badly sick with cold on that day. In fact, he was sneezing while approaching the stop sign. As a result he involuntary pushed the brakes very hard. Therefore we can assume that the deceleration was close to maximum possible for a car, which is of the order of 10 m/s 2 = 22.36 mph/s. We will thus use a [Annotation 5] 0 = 10 m/s 2 . Substituting these values of a 0 , x p , and x f into Eq. (8) inverted for t,

t =

r

2x a 0

, (14)

we obtain

t p = 1.31 s, (15) t f = 0.45 s. (16)

The full durations of the partial and full obstructions are then just double these times.

Next, we are interested in time t 0 at which the angular speed of C 1 observed by O without any obstructions goes over its maxima, as in Fig. 3. The easiest way to ﬁnd t 0 is to recall that the value of the ﬁrst derivative of the angular speed at t 0 is zero,

dω dt = ˙ω(t 0 ) = 0. (17)

To ﬁnd ˙ω(t) we just diﬀerentiate Eq. (13) using the standard diﬀerentiation rules, which yield

˙ω(t) = 4 a 0 r 0

1 − 3 4



a 0 r 0

 2

t 4 

1 + 1 4



a 0 r 0

 2

t 4

 2 . (18)

This function is zero only when the numerator is zero, so that the root of Eq. (17) is

t 0 = 4

r

4 3

r

r 0 a 0

. (19)

Substituting the values of a 0 = 10 m/s 2 and r 0 = 10 m in this expression, we obtain

t 0 = 1.07 s. (20)

We thus conclude that time t 0 lies between t f and t p ,

t f < t 0 < t p , (21)

and that diﬀerences between all these times is actually quite small, compare Eqs. (15,16,20).

These ﬁndings mean that the angular speed of C 1 as observed by O went over its maxima when the O’s view of C 1 was partially obstructed by C 2 , and very close in time to the full obstruction. In lack of complete information, O interpolated the available data, i.e., the data for times t > t 0 ≈ t f ≈ t p , using the simplest and physiologically explainable linear interpolation, i.e., by connecting the boundaries of available data by a linear function. The result of this interpolation is shown by the dashed curve in Fig. 5. It is remarkably similar to the curve showing the angular speed of a hypothetical object moving at constant speed v 0 = 8 m/s ≈ 18 mph.

V. CONCLUSION

In summary, police oﬃcer O made a mistake, confusing the real spacetime trajectory of car C 1 —which moved at approximately constant linear deceleration, came to a

#### Page 4

4

−10 −8 −6 −4 −2 0 2 4 6 8 10 0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

t (seconds)

ω(t) (radians per second)

C 1 ’s speed O’s perception v 0 = 8 m/s

t f

t′

t p

FIG. 5: The real angular speed of C 1 is shown by the blue solid curve. The O’s interpolation is the dashed red curve. This curve is remarkably similar to the red solid curve, showing the angular speed of a hypothetical object moving at constant linear speed v 0 = 8 m/s = 17.90 mph.

complete stop at the stop sign, and then started moving again with the same acceleration, the blue solid line in Fig. 5—for a trajectory of a hypothetical object moving at approximately constant linear speed without stopping at the stop sign, the red solid line in the same ﬁgure. However, this mistake is fully justiﬁed, and it was made possible by a combination of the following three factors:

1. O was not measuring the linear speed of C 1 by any special devices; instead, he was estimating the visual angular speed of C 1 ;

2. the linear deceleration and acceleration of C 1 were relatively high; and

3. the O’s view of C [Annotation 6] 1 was brieﬂy obstructed by another car C 2 around time t = 0.

As a result of this unfortunate coincidence, the O’s perception of reality did not properly reﬂect reality.

---

Source: [Fermat's Library](https://fermatslibrary.com/) — https://fermatslibrary.com/s/the-proof-of-innocence. More annotated papers: https://fermatslibrary.com/llms.txt
