Sıkısınca

Last week saw the popularization of some older news about a misunderstanding, prompted by an error caused by technological limitations of mobile phones, that resulted in two deaths and three imprisonments. The whole sad story took place in Turkey. You can read the original story in the Turkish language Hürriyet.

Basically, Emine and her husband Ramazan Çalçoban has recently been separated and were feuding daily on their mobile phones and over SMS text messages. At one point, Ramazan sent a message saying, “you change the subject every time you get backed into a corner.” The word for “backed into a corner” is sıkısınca. Notice the lack of dots on the i’s in the word. The very similar sikisince — spelled with dots — means “getting fucked.” Ramazan’s mobile phone could not produce the “closed” dotless ı so it he wrote the word with dots and sent it anyway. Reading quickly, Emine misinterpreted the message thinking that Ramazan was saying, “you change the subject every time they are fucking you.” Emine showed the message to her father and sisters who, outraged that Ramazan was calling Emine a whore, attacked Ramazan with knifes when he showed up at the house later. In the fight, Ramazan stabbed Emine back and she later died of bleeding. Ramazan committed suicide in jail and Emine’s father and sisters were all arrested.

This is certainly the gravest example of a revealing error I’ve looked at yet and it stands as an example of the degree to which tiny technological constraints can have profound unanticipated consequences. In this case, the lack of technological support for characters used in Turkish resulted in the creation of text that was deeply, even fatally, ambiguous.

Of course, many messages sent with SMS, email, or chat systems are ambiguous. Emoticons are an example of a tool that society has created to disambiguate phrases in text-based chatting and their popularity can tell us a lot about what purely word-based chatting fails to convey easily. For example, a particular emoticon might be employed to help convey sarcasm in a statement that would have been obvious through tone of voice. One can think of verbal communication as happening over many channels (e.g., voice, facial expressions, posture, words, etc). Text-based communication technologies provide certain new channels that may be good at conveying certain types of messages but not others. Emoticons, and accents or diacritical marks for that matter, are an attempt to concisely provide missing information that might be taken for granted in spoken conversation.

Any communication technology conveys certain things better than others. Each provides a set of channels that convey some types of messages but not others. The result of a shift toward many technologies is lost channels and an increase in ambiguity.

In spoken Turkish, the open and closed i sounds are easily distinguishable. In written communication, however, things become more difficult. Some writing system are better at conveying these tonal differences. Hebrew, for example, historically contained no vowels at all! And yet, the consequence of not conveying these differences can be profound. As a result, Turkish speakers frequently use diacritics and the open and closed i notation to disambiguate phrases like the one at the center of this saga. Unfortunately the open and closed i technology is not always available to communicators. Notably, it was not available on Ramazan’s mobile phone.

People in Turkey have ways of coping with the lack of accents and diacritical marks. For example, some people would choose to write sıkısınca as SIKISINCA because the capital I in the Roman alphabet has no dot. Emoticons are similar in that they are created by users to work around limitations of the system to convey certain messages and to disambiguate others. In these ways and others, users of technologies find creative ways of working with and around the limitations and affordances imposed on then.

With time though, the users of emoticons and all-caps Turkish words stop seeing and thinking about the limitations that these tactics expose in their technology. In fact, it is only through errors that these limitations become familiar again. While we cannot undo the damage done by Ramazan, Emine and her family, we can “learn from their errors” and reflect on the ways that the limits imposed by our communication technology frames and affects our communications and our lives.

Interpolation

One set of errors that almost everyone has seen — even if they don’t know it — involve the failure of a very common process in computer programming called interpolation. While they look quite different, both of the following errors — each taken from the Daily WTF’s Error’d Series — represent an error whose source would be obvious to most computer programmers.

You Saved a total of {@Total-Tkt-Discount} off list prices.

The term interpolation, of course, is not unique to programmers. It is a much older term that was historically used to describe errors in hand-copied documents. Interpolation in a manuscript refers to text not written by an original author that was inserted over time — either through nefarious adulteration or just by accident. As texts were copied by hand, this type of error ended up happening quite frequently! In its article on manuscript interpolation, Wikipedia describes one way that these errors occurred:

If a scribe made an error when copying a text and omitted some lines, he would have tended to include the omitted material in the margin. However, margin notes made by readers are present in almost all manuscripts. Therefore a different scribe seeking to produce a copy of the manuscript perhaps many years later could find it very difficult to determine whether a margin note was an omission made by the previous scribe (which should be included in the text), or simply a note made by a reader (which should be ignored or kept in the margin).

But while manuscript interpolation described a type of error, interpolation in computer programming refers to a type of text swapping that is fully intentional.

Computer interpolation happens when computers create customized and contextualized messages — and they do so constantly. Whereas a newspaper or a book will be the same for each of its readers, computers create custom pages designed for each user — you see these all the time as most messages that computers print are, in some way, dynamic. In many cases, these dynamic messages are created through a process called string or variable interpolation. For those who are unfamiliar with the process, an explanation of the errors above can reveal the details.

In the first example, the receipt read (emphasis mine):

You Saved a total of {@Total-Tkt-Discount} off list prices.

In fact, the computer is supposed to swap out the phrase {@Total-Tkt-Discount} for the value of a variable called Total-Tkt-Discount. The {@SOMETHING} syntax is one programming language’s way of signifying to the computer, “take the variable called SOMETHING and use its value in this string instead of the everything between (and including) the curly braces.” Of course, something didn’t quite work right and the unprocessed — or uninterpolated — text was spit out instead. With this error, the computer program that is supposed to be computing our ticket price was revealed. Additionally, we have a glimpse into the program, its variable names, and even its programming language.

The second error from a (not very helpful) dialog box in Mozilla Firefox is a more complicated but fundamentally similar example (emphasis mine):

The file “#3” is of type #2 (#1), and #4 does not know how to handle this file type.

The numbers, in this case, reflect a series of variables. The dialog is supposed to be passed a list of values including the file name (#3), the file type (#2 and #1), and the name of the program that is trying to open it (#4). This list is supposed to be swapped in from placeholder values — interpolated — before any user sees it. Again, something went wrong here and a user was presented with the empty template that only the programmer and the program are ever supposed to see.

Nearly every message a computer or a computerized system presents us will be processed and interpolated in this way. In this sense, computer programs act as powerful intermediaries processing and displaying data. Perhaps more importantly, interpolation reveals just how limited computers’ expression really is. These messages are not more complicated than simple fill-in-the-blank messages. Simple as they may be, they are entirely typical of the way that computers communicate with us.

From a user’s perspective, it’s easy to imagine sophisticated systems creating and presenting highly dynamic messages to us — or to simply not think about it at all. In reality, few computer programs’ ability to communicate with us is more sophisticated than a game of Mad Libs. The simplicity of these systems, the limitations that they impose on what computers can and can’t say, and the limitations they place on we can and can’t say with computers, are revealed through these simple, common, interpolation errors. To understand all of this, we need only recognize these errors and reflect on what they might reveal.