From October 2016 to March 2017 the team is joined by Guest Kats Rosie Burbidge and Eibhlin Vardy, and by InternKats Verónica Rodríguez Arguijo, Tian Lu and Hayleigh Bosher.

Friday, 14 October 2011

The daylight-saving data dispute

Astronomical clock at St Mark's, Venice
Astrologers believe that the exact position of celestial bodies at a given instant, such as when a baby is born, is a predictor of the future; while computer scientists believe (somewhat more rationally, Merpel suspects) that unless computers in a network agree what time it is, communications are likely to fail in spectacular and unpredictable ways. Both groups therefore have a shared interest in knowing precisely how the local time - as shown by the clock hanging on the wall of a hospital delivery room, or as displayed on a computer monitor - relates to Coordinated Universal Time (which is, for our purposes, the same as Greenwich Mean Time or GMT).

So why has a company which sells astrology tools and books sued two very distinguished and serious researchers from the United States, who have spent many years documenting the minutiae of daylight savings time transitions in every corner of the globe? 

The researchers are Arthur David Olson, who works for the US National Institute of Health, and Paul Eggert, a lecturer in computer science from UCLA. The results of their work, documenting the exact dates and times on every occasion when local time was adjusted almost everywhere in the world since the late 19th century, are published in an open source database known as the "TZ database". The astrology company says that they own copyright in an earlier collection of data on time changes, which they say has been infringed by Olson and Eggert.

The TZ Database
This database (also called the zoneinfo database) is the largest and most authoritative collection of facts in existence about daylight saving time changes, time zones, and time adjustments. And not just daylight savings: time is adjusted for other reasons. Subjugated countries are sometimes forced to synch their clocks to the local time of a conquering nation. The USA had year-round DST in the 1942-1945 period to conserve energy. Whatever the reason, it's in the TZ database.

The database is replete with comments, quotations and quirky historical facts justifying why the authors (with the assistance of many contributors from around the world) have concluded that in a given location, on a given date and time, the local time was adjusted in a particular way relative to GMT. *

In addition to the human-readable comments in the database, the information is expressed as a set of rules which a computer can interpret. For example, in the case of the year-round daylight savings time in the USA during WWII, there is a rule which tells computers: in all of the time zones for the United States, during the period from 02:00 on February 9, 1942 to 02:00 on September 30, 1945, apply an adjustment of + 1 hour relative to standard time.

Unix-based computers (i.e. most of the web servers which delivered this blog posting or email to you) rely on the TZ database to translate between the idealised time of atomic clocks and the actual local time at their location. The purpose of the TZ database, therefore, is to provide an exhaustive repository of rules which computers can use to calculate the local time at every location on earth throughout history and into the future. They can use the rules to convert a local time into a unique, objectively agreed number, which is the precise number of seconds until (or since) the stroke of midnight UTC on January 1 1970. It works in both directions: the unique number ascribed to a given moment according to UTC can be converted back into the "wall clock time" as it would have appeared at any location on earth at that instant. So the TZ database lets computers agree on when events happened (or will happen). This is critical to the operation not only of the Internet, but also of many computer programs and operations.

The Lawsuit

The problem with putting the clocks
back by an hour each year is that humans 

think this justifies  withholding a 
cat's dinner by an hour as well
So that explains the TZ database, but not why the astrology publisher, Astrolabe Inc., objects to its continued publication. Some years ago Astrolabe Inc. bought up the rights to one of the foremost collections of data on time adjustments, known as the ACS Atlas. This data set is widely used by astrologers and (it turns out) by Olson and Eggert who relied on the ACS Atlas as one of their primary sources.

The issue is summarised in the Complaint filed by Astrolabe Inc. as follows:
Pursuant to a written agreement, Astrolabe is the copyright assignee of [...] the “ACS Atlas,” consisting of both the “ACS International Atlas,” and the “ACS American Atlas,” in the form of computer software program(s) and/or data bases, and in the form of electronic output and future electronic media from said programs [hereinafter “the Works”].
These atlases set forth interpretations of historical time zone information pertaining to innumerable locations throughout the world, based upon the compilation of historical research and documentation regarding applicable time zones officially and/or in actuality in effect, given the actual latitude and longitudes of specific locations throughout the world.
Upon information and belief, defendants Olson and Eggert have unlawfully reproduced the Works, in violation of the Copyright Protection Act, without proper permission and/or authorization from the copyright holder, and without paying royalties due and payable to the copyright holder and/or its assignee, Astrolabe, in computer software format.
The Complaint goes on to note that the TZ database, published by the defendants, is "replete with references to the fact that the source for this information is, indeed, the ACS Atlas." 

Of course, the fact that a reference work was used as a source does not of itself mean that copyright has been infringed, and there is probably a significant issue over exactly what Olson and Eggert have taken - is it copyrightable expression, mere facts, or both? We don't know enough to answer these questions, but it promises to be an interesting case, should it go to trial. Some of the issues which may arise (no doubt readers can think of others) are:
  • Does US copyright law protect collections of historical data such as when time changes occurred in various regions?
  • How does this fit with the Feist case in which the US Supreme Court refused copyright protection to telephone directories?
  • If there is copyright in the ACS Atlas, is it infringed when used as a primary source as Olson and Eggert have done? Have they gone beyond a permissible reuse of facts?
  • Is there a fair use defence open to Olson and Eggert?
  • Given that the TZ Database project began in 1986, is there any estoppel or acquiescence that can be argued?
  • If Astrolabe Inc. ultimately prevail, what does this mean for all of the computer systems and programs which rely on the future maintenance of the TZ database for their continued operation?
  • Perhaps most intriguingly, have Astrolabe Inc. consulted the positions of the heavenly bodies to predict their prospects of success before the courts? And if so, should this information be discoverable?
Mr Olson has posted a brief message on the site where he hosted the TZ database as follows:
A civil suit was filed on September 30 in federal court in Boston; I'm a defendant; the case involves the time zone database.The ftp server at has been shut down. The mailing list will be shut down after this message. Electronic mail can be sent to me at arthurdavidolson <at> I hope there will be better news shortly.
In response, several sites outside the USA have made copies of the database available, so you can rest assured the internet is not going to break down just yet.

The time ball mentioned by Leopold Bloom
can be seen on the corner of the roof
of Dublin's Ballast Office, after having dropped.
* This Kat looked up his own time zone and it seems that between 1880 and 1916, Dublin Mean Time was calculated not on the basis of the actual latitude of the city, but rather with reference to Dunsink Observatory, 8 km northwest (meaning that a clock in Dublin city centre would strike noon about 20 seconds after the sun had passed its highest point overhead). The database includes a quote from James Joyce, ever a scrupulous observer of Dublin life, in support of this finding. In Ulysses, Leopold Bloom notes the hour striking from the dropping of a time ball, and he says: "Timeball on the Ballast Office is down. Dunsink Time."


Francis Davey said...

The need for agreement between computers about time is a Stephen Fry propagated myth:

I shudder to think what would happen if it were true. Even using good time synchronisation protocols in practice lots of computers drift quite a bit. Well designed network protocols don't assume that the ends are able to synchronise their clocks and rightly so.

Differences in time stamps can cause problems - eg for distributed file systems where conflicts of edits are resolved via the time stamp - but that's a secondary not a primary problem (its at a much higher layer).

TZ has nothing to do with this - as you sort of explain - since it maps time (usually something like Universal Time) to local time conventions. Most modern computers don't think in local time (there were exceptions in the past - different flavours of windows disagreed about how to do it which was fun in the 90's) so TZ would have nothing to do with synchrony anyway.

Of course that has nothing to do with IP. Just an aside.

Anonymous said...

There is a lot of nonsense propogated by Stephen Fry as fact. The loveys (actors) of the world believe he is the smartest person on the planet, but they have no understanding of the concept of relativity. My goldfish has a better understanding of science and he floats on his side.

Anonymous said...

As to the copyright isue, at least in the US, there is no copyrihgt available on "pure data."

Francis Davey said...

ICANN have stepped in:

Subscribe to the IPKat's posts by email here

Just pop your email address into the box and click 'Subscribe':