Scientific Time Standards

Part II Certainty in Information Systems

CHAPTER 5: COMPUTATIONAL TIMEKEEPING

5.2 Scientific Time Standards

The development and promulgation of scientific time standards predates the computational timekeeping

enterprise, but is in retrospect its necessary precondition. Practices of computational timekeeping don’t

5_{This is memorialized in the}_tz_{database, we’ll see below, as the first entry in most US zone definitions used to} generate human-readable datetimes in computers.

6_{On the topic of Daylight Saving Time see Michael Downing,}_{Spring Forward: The Annual Madness of Daylight}

Saving(Berkeley, Ca: Counterpoint, 2009); David Prerau,Seize the Daylight: The Curious and Contentious Story of Daylight Saving Time(New York, NY: Avalon Publishing Group, 2005).

7_{For more on the development of standard time globally, Bartky has produced perhaps the most compelling and} scholarly accounts of the development of standard time movements in the US and Europe. Ian R Bartky,One Time Fits All: Campaigns for Global Uniformity(Stanford, CA: Stanford University Press, 2007).

always merely recapitulate scientific standard time, but practically all methods in modern use utilize them

as a kind of shared backbone, the terms in which their work is stated. So, as a prelude to considering major

components of computational timekeeping, this section will describe major components of scientific

standard time, and detail the methods of its promulgation.

5.2.1 Empirical Time: TAI & UT1

UTC is a boundary object produced by the promulgators of scientific standard time, professional

metrologists at the BIPM. The standards utilized in its creation, TAI and UT1, are rarely considered by

those not involved in metrology or the production of standards and, consequently, don’t carry UTC’s

negotiated meanings in political, computational, and quotidian contexts. In the terms of the present

investigation, UTC is a place to be moved on from, designed for the enactment of certainty. Like all

promulgations, TAI and UT1 instantiate a technique of forgetting which collapses the complexity of their

creation into publications likeCircular Tor electromagnetic signals of precise frequencies.

Produced as

it is by the scientific community, this technique of forgetting bottoms out in measurements.

5.2.2 Calculating Duration with TAI

TAI is constructed through the incorporation of empirical measurements of duration from around 100

time laboratories collectively containing several hundred high-precision clocks around the globe (a map

is provided in Figure 6.2 (page 158)). According to the BIPM,

The long-term stability of TAI is assured by weighting the participating clocks. The scale unit of TAI is kept as close as possible to the SI second by using data from those national laboratories which maintain the best primary caesium standards.9

The SI standard second (discussed further on page 159) supports the act of incorrigible declaration of

this unit with one of reconcilable measurement. Scientific techniques of forgetting govern this change,

which evinces the metrological instinct to base scientific units on as few, and as accurate, measurements

as possible.

8_{B G Quinn and E J Hannan,}_{The Estimation and Tracking of Frequency}_{(Cambridge, UK: Cambridge University} Press, 2001) provides a technical introduction to the measurement of frequency.

9_{“International Atomic Time (TAI),” BIPM, accessed October 9, 2019,} _{https://www.bipm.org/en/} bipm-services/timescales/tai.html

5.2.3 Locating the Earth in Spacetime with UT1

Duration, however is insufficient for the provision of time. While seconds are defined precisely as a

specific duration, other units of time important to daily life are not evenly or reliably composed of a fixed

number of seconds. A day, the duration of a single revolution of the earth, traditionally held to consist

of 86400 seconds, does not exactly equal this duration (this is a more obscure corollary to the fact that

years are not composed of exactly 365 days). What’s more, the precise duration of a day varies over time

with the angular momentum of the earth, which is currently slowing. Cultural practices that rely on the

assumption of even divisibility of days into seconds and years into days are enabled by the invisible

efforts of metrologists and the faithful enactment of their promulgations by computational timekeeping

devices.

The measurement of the earth’s angular momentum (i.e., how fast it is turning on its axis) requires

establishing an invariant frame of reference. This frame of reference is established by the International

Celestial Reference Frame (ICRF), itself produced by the astronomical observation of the quasars and

other stable extragalactic sources of radiation.

In short, UT1 precisely tracks the orientation of earth so

as to determine when a full revolution has occurred.

5.2.4 Coordinating Duration and Location: UTC

Given that the the number of SI seconds isn’t evenly divisible into the length of a day, a process of

coordinating these two values is required to maintain a calendrical system which declares that there are

86,400 seconds in a day. This is accomplished with the periodic insertion of ‘leap seconds,’ designed to

keep TAI and UT1 within±0.9sof each other, which produces UTC. The BIPM describes UTC on its

website as follows.

Coordinated Universal Time (UTC), maintained by the BIPM, is the time scale that forms the basis for the coordinated dissemination of standard frequencies and time signals. The UTC scale is adjusted by the insertion of leap seconds to ensure approximate agreement with the time derived from the

10_{Chopo Ma and Martine Feissel, eds.,}_{Definition and realization of the International Celestial Reference System}

by VLBI astrometry of extragalactic objects, technical report {IERS Technical Note 23} (Central Bureau of IERS – Observatoire de Paris, 1997). See also “The International Celestial Reference Frame (ICRF),” ICRF, accessed

October 9th, 2019,https://www.iers.org/IERS/EN/DataProducts/ICRF/ICRF/icrf.html.

11_{This improves upon earlier efforts from the 18}th_{and 19}th_{centuries to calculate the duration of a second using} observations of the Moon to establish a reference frame.

rotation of the Earth. These leap seconds are inserted on the advice of the International Earth Rotation and Reference Systems Service (IERS).

Physical realizations of UTC – named UTC(k) – are maintained in national metrology institutes or observatories contributing with their clock data to the BIPM.12

The earth’s rotation is currently slowing, so all 38 leap seconds to date have been positive. If metrologists

are still actively maintaining these standards when the earth’s rotation speeds up, negative leap seconds

will be added to keep TAI and UT1 within±0.9s.

5.2.5 The Current State of Universal Standard Time

Standard time, which could be seen as a set of synchronization behaviors, grew into the various officially

promulgated Universal Time standards. The availability of scientifically defined universal time formed a

substrate for a proliferating array of administratively declared time zones, which encode synchronization

behaviors but have a social reality beyond them. Figure 5.2 on the following page shows the resulting

complexity.

This complexity has been enabled by both the scientific definition of standard time and techniques

of computational timekeeping, particularly thetzdatabase (discussed in Section 6.2). Standard time

was historically envisioned as neat lines drawn along the map at regular intervals according to longitude,

so that all world timezones would be exactly 1 hour from each other, and, on average, equidistant from

their local solar times. But the substrate of universal standard time when combined with computational

timekeeping has enabled a practically endless array of variations from this ideal, which are plainly visible

by the interruptions and articulations of the straight vertical lines in Figure 5.2.

With a history of standard time and a basic description of modern universal time standards completed,

we can now turn to the components of computational time. The adoption of UTC in 1972 was portentous

for the development of computational timekeeping. Computational timekeeping, and networked comput-

ing more broadly, has in turn influenced and accelerated the adoption of standardized time to its current

near-ubiquitous state.

12_{“Approximation} _to _UTC”, _BIPM, _accessed _October _9, _2019,_{https://www.bipm.org/en/}