# A universal law of procrastination

> 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/a-universal-law-of-procrastination

- **Authors:** Tomasz Durakiewicz
- **Published:** 2016
- **Topic:** Biology
- **Original:** https://s3-us-west-1.amazonaws.com/fermatslibrary/uploads%2F1667137816025-8vom7rogj5-19929e679b05ca1b595d44b020759c0e%2Fdurakiewicz2016.pdf
- **Pages:** 2
- **Annotations:** 4 comments in 3 threads
- **Featured:** Fermat's Library Journal Club
- **Annotated edition:** https://fermatslibrary.com/s/a-universal-law-of-procrastination

## Annotations

### Annotation 1 (page 1)

Beside the passage: “hroughout our lives we all are under pressure to deliver on deadlines. Yet we often have a substantial time window in which to complete the given tasks—for example, term papers, book chapters, tax returns. When during such a window do we deliver? And to what extent do we procrastinate?”

**Annotation** by Luís Batalha (2022-10-31):

> A good example of this behavior can be seen in the number of paper [submissions for Neural Information Processing Systems (NeurIPS)](https://homepages.inf.ed.ac.uk/imurray2/submission_times/), one of the most prestigious machine learning and computational neuroscience conferences in the world.
>
> ![](https://homepages.inf.ed.ac.uk/imurray2/submission_times/subs.png)

### Annotation 2 (page 1)

Beside the passage: “PROPOSED MODIFIED HYPERBOLIC function reflecting a universal law of p r o c r a s t i n a t i o n . T h e r e d l i n e r e p r e s e n t s the full form of the law, which illustrates the hyperbolic scaling, N(t) is the number of submissions received by day t, M is the final number of proposals submitted, and D is the …”

**Annotation** by Luís Batalha (2022-10-31):

> Recent research has shown that a hyperbolic function provides a more accurate description of the deadline rush than an exponential.
>
> ![](https://i.imgur.com/9I3Sd3p_d.webp?maxwidth=760&fidelity=grand)
>
>  The difference between hyperbolic and exponential discounting is not trivial: the different shapes have implications for the consistency of preferences.
>
>  1) If preferences follow the exponential curve, they are consistent over time (i.e., stationary)
>
>  2) If preferences follow the hyperbolic curve, they do not have to be consistent over time and preference reversal can be expected: People prefer a larger but later reward over a smaller but sooner reward when both outcomes are available in the far future (the smaller reward still being available relatively sooner than the larger reward); however, they prefer the smaller but sooner reward over the larger but later reward when some time has passed and the smaller but sooner reward is immediately available. Thus, their preferences reverse by shifting toward the smaller but sooner reward when time goes by.
>
> An example of preferences reversing is a person that prioritizes her tasks on Friday for the coming week based on the hyperbolic discounted value of the tasks. On Monday, even though the plan is to work on some important project
> involving a larger but later reward, an e-mail arrives on the inbox and she changes her mind and works to get smaller but sooner rewards and does not work on the important task with the larger but later reward. This is why important but nonurgent tasks are often neglected, unlike urgent but unimportant tasks, which is a topic frequently mentioned in time management self-help books!

### Annotation 3 (page 1)

Beside the passage: “PROPOSED MODIFIED HYPERBOLIC function reflecting a universal law of p r o c r a s t i n a t i o n . T h e r e d l i n e r e p r e s e n t s the full form of the law, which illustrates the hyperbolic scaling, N(t) is the number of submissions received by day t, M is the final number of proposals submitted, and D is the …”

**Annotation** by Luís Batalha (2022-10-31):

> The core of the behavioral decision-making approach to time management
> is that people discount the value of future outcomes. Time discounting means people prefer to receive a reward sooner rather than later, because the present subjective value placed on a delayed reward decreases as the delay to that reward increases.
>
> People often have to choose between short-term and long-term rewards, and time discounting has major implications for these decisions, for instance: a drug addict chooses between the positive effects of taking a drug and the negative
> effects of drug use, such as contagious diseases and loss of employment, in the
> future.
>
> Time discounting can have important implications for decisions on how people spend their time. In particular, time discounting can be used to explain the rush before a deadline: it is not very likely that people will work now on a project with a deadline in the far future because people now discount the outcomes that are the results of this project.

**Reply** by Rahul Ranjan (2023-12-20):

> Is there a strategy to avoid discounting? How can we reverse the psychology of long-term gain to short-term?

## Full text

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

#### Page 1

FEBRUARY 2016 | PHYSICS TODAY 11

clearly shows the conducting properties of a semiconductor at various conditions. By the last lecture in that section, the students understand the principle.

Using new technologies. With the ever-increasing availability of online education resources, including open course - ware, online courses, and discussion forums, choosing the best resources is becoming a new challenge for instructors. However, electronic resources provide new tools and opportunities to increase students’ conceptual understanding.

I have adopted a lot of electronic resources in teaching both Thermal Phys - ics and Electromagnetism. For example, Bose–Einstein condensation and superﬂuidity are subtle processes that can be

hard to understand, and few laboratory teams have achieved those results experimentally. Videos played in the classroom allow students to watch the phase transition when liquid helium-4 is cooled to a temperature of 2.17 K and becomes a superﬂuid.

References 1. J. B. Biggs, Student Approaches to Learning

and Studying, Australian Council for Educational Research (1987). 2. K. Aikawa et al., Science 345, 1484 (2014). 3. E. B. Li et al., Nat. Commun. 5, 3225 (2014).

Enbang Li (enbang@uow.edu.au) University of Wollongong

Wollongong, Australia

A universal law of procrastination

T

hroughout our lives we all are under pressure to deliver on deadlines. Yet we often have a substantial time window in which to complete the given tasks—for example, term papers, book chapters, tax returns. When during such a window do we deliver? And to what [Annotation 1] extent do we procrastinate?

Increasing pressure forces us to ﬁnalize tasks. The closer we are to the deadline, the higher the pressure. To quantify things a bit, we could assume that deadline pressure is inversely proportional to remaining time. Such scaling behavior is well known in physics; for example, it describes how the electrostatic potential energy of a point charge depends on distance from the charge, or how the gravitational potential energy of a system of masses scales with distance between the masses.

The scales look like 1/r, where r can be distance, remaining time, or some other factor. The corresponding mathematical function is a hyperbola. To give an example of the scaling using arbitrary and dimensionless units, if a deadline is 100 days away, the pressure to meet it is just 0.01. But if the deadline is tomorrow, the pressure shoots up to 1.

As it turns out, that scaling may be universal when it comes to human behavior and perhaps reﬂects a universal law of procrastination (ULP). Cornelius König and Martin Kleinmann reported a hyperbolic “deadline rush” for a small set of fewer than 30 students prepar-

ing for exams over a 21-day period. 1 As a program director at NSF, I was impressed by the tremendous increase in the number of proposals routinely submi"ed right before deadline as compared with those turned in earlier in the submission window. In analyzing large data sets from 10 annual submission

M

0

Nt M D t C M D C()= /( − + )− /( + )

0 D

SUBMITTED PROPOSALS

TIME

PROPOSED MODIFIED HYPERBOLIC function reflecting a universal law of p r o c r a s t i n a t i o n . T h e r e d l i n e r e p r e s e n t s the full form of the law, which illustrates the hyperbolic scaling, N(t) is the number of submissions received by day t, M is the final number of proposals submitted, and D is the number of days in the submission window. According to the universal law of procrastination, N(t) = M/(D – t + C) − M/(D + C), where C = ½ (√D · √4 + D – D) helps improve [Annotation 2] the fit for small values of t. The second term in the right-hand side of the equation e n s u r e s t h a t N(t) = 0 at t = 0. Asymptote (not shown) is located at t = D + C. [Annotation 3]

#### Page 2

12 PHYSICS TODAY | FEBRUARY 2016

READERS’ FORUM

windows of greater than 60 days with submissions from more than 1000 proposers annually, I have found that the data adhere to a modiﬁed hyperbolic function, as plo"ed in the ﬁgure. The model is simple, which means it omits factors such as delays introduced by the universities’ sponsored research oﬃces. Nevertheless, the procrastination behavior is predicted quite well, and without any ﬁ"ing parameters.

As is shown in the ﬁgure, the ULP curve provides a simple means of predicting the impact of proposal pressure and of estimating the number of proposals expected as a function of remaining time to deadline. Practical concerns for a receiving institution include how to handle the number of proposals received on the deadline date and whether that load will overtax or crash the existing computer infrastructure.

Bear in mind, though, that hyperbolic functions diverge to inﬁnity at the asymptote. To procrastinating submi"ers, the most critical issue is that by waiting until the deadline or close to it, they eliminate the time needed for identifying and correcting errors that could make their proposal ineligible for consideration.

I appreciate helpful discussions with Andy Lovinger at NSF and the encouragement of Cornelius König of Saarland University. Any opinions expressed in this material are mine and do not necessarily reﬂect the views of NSF.

Reference 1. C. J. König, M. Kleinmann, J. Psychol. 139,

33 (2005).

Tomasz Durakiewicz

(tomasz@lanl.gov) National Science Foundation

Arlington, Virginia

Pictures of

climate change

S

pencer Weart’s article on climate impacts (P HYSICS TODAY, September 2015, page 46) describes the sociology of how opinion has evolved on anthropogenic change, but it says li"le about the opinion’s scientiﬁc content. It is remarkable that the scientiﬁc giant in this ﬁeld, Svante Arrhenius (1859–1927), without knowledge of the Planck func-

tion—much less the quantum mechanics of molecular opacity or computer codes—made predictions of climate sensitivity that are within a factor of two or three of modern estimates. Was that a lucky guess, or is the phenomenon so robust that even the crudest estimates are almost as good as the most sophisticated?

Weart describes, but does not explain, how the consensus about the eﬀects of climate change has shi(ed from equanimity to fear and trembling that a great disaster will ensue. Is climate change a phenomenon to be observed, like the weather? Is it of direct concern mostly to farmers? Or is it a problem to be solved, and if so, how urgently? The shi( is a sociological phenomenon that calls for explanation, but not by physicists.

The physical principles have long been known, and François Massonnet’s Commentary in the same issue (page 8) explains that even our present understanding and computational capabilities are not suﬃcient to predict regional effects such as droughts and ﬂoods. The fact that multiphysics codes—which combine multiple models to simulate complex phenomena—could not predict the failure of National Ignition Facility targets should make us skeptical of their power to predict any complex phenomenon, and climate is more complicated than a laser target.

Jonathan Katz (katz@wuphys.wustl.edu) Washington University in St. Louis

St. Louis, Missouri

!!!

H

aving formerly worked for the National Weather Service for 40 years, including assignments at the National Severe Storms Forecast Center and various ﬁeld forecast oﬃces, I was struck by the images in Spencer Weart’s article “Climate change impacts: The growth of understanding.” I thought it was interesting that the editors chose to illustrate the article with several weather-disaster photos.

The cover photo shows ﬂooding of small ﬁelds lined with palms and other tropical fauna. Other photos show drought and floodwaters extending halfway up storefront shops.

The inference, I suppose, is that climate change caused those weather disasters, despite the author’s stating he was unable “to present a convincing case, based on

logic and observations, of why anyone should believe the consensus statements” about climate change impacts.

Those photographs perhaps make it more pleasing visually to leaf through a publication, but their inclusion only perpetuates the myth that individual storms are the result of climate change. For example, the vast majority of the ﬂooding shown in the Hurricane Sandy photo was due to the storm surge that typically accompanies hurricanes. The track of Hurricane Sandy was an outlier in the data set. The unusual ﬂooding can be explained entirely by storm dynamics over the ocean. A sea-level rise of several inches due to ice melt would not by itself cause 20- to 25-foot storm surges.

John T. Curran (jtcurran41@gmail.com)

Carmel, Indiana

‣ Weart replies: Jonathan Katz worries about the validity of computer studies of projected impacts of climate change. And John Curran notes that illustrations to my article show particular events, which computer studies indeed have difﬁculty a"ributing individually to climate change. I apologize if any reader jumped to the conclusion that a speciﬁc a"ribution was intended. I wanted only to illustrate the subject of the article—namely, impacts in general. Still, part of the sealevel rise of the past century is reliably a"ributed to global warming, and the rise did extend the area of Sandy’s inundation. And a peer-reviewed study has reported that global warming did contribute to the Texas drought that was illustrated.

For reasons of length I had to leave out the interesting story of a"ribution studies of particular impacts; for a sketch and references see h"p://www .aip.org/history/climate/impacts.htm. Researchers have labored for decades to test computer models against observations, and the matches have been good although imperfect. Anyway, it is not computers but simply the thermal expansion of water and the visible decay of ice sheets that support expectations of further sea-level rise if greenhouse gas emissions continue. Other serious impacts have already been observed in weather statistics, including global intensiﬁcation of heat waves and of extreme precipitation events.

Finally, Curran misunderstands a

---

Source: [Fermat's Library](https://fermatslibrary.com/) — https://fermatslibrary.com/s/a-universal-law-of-procrastination. More annotated papers: https://fermatslibrary.com/llms.txt
