Theory of the Folder "incoming"

So you sent someone an email, terribly important; How to wait soon? Or, conversely, you are asked about something important with a letter, and you saw a message, but you are not ready to react on the move. Will it be impolite to answer tomorrow? It seems that yes: half the letters are answered in the first 47 minutes. Moreover, if your interlocutor writes from a smartphone, then with a probability of 50 percent the answer will come in 28 minutes or earlier.

A team of scientists from the University of South California and Yahoo Labs , the Yahoo Corporation, analyzed the largest email array in history - 16 billion letters in the boxes of two million users (those who gave Yahoo consent to this). Of course, no one read the letters themselves - moreover, the scientists did not have direct access to their text; It was a typical computer experiment with “big data”.

Imagine that all your correspondence went through the hands of a paranoic censor, which covered each letter in black by black, and the fields “who” and “from whom” were sealed with tags with conditional ordinal numbers-say, the Sh-451 and F-255 (but indicated the gender and age nearby). The authors say: the meager information that remains with a margin is enough to predict your actions in the future. Guess, for example, how much time will be pulled with the following answer. What will be his message - short or long. And whether he will put the next letter from the pc in the dialogue (the algorithm confidently guesses in two cases out of three).

Avoid invested files when you want a quick response.

The forecast is based on statistics: before the final replica in the chain of letters, people for some reason withstand a particularly long pause. There were dozens of such patterns.

Answers to letters sent on the weekend are especially short. It is not enough to count on a quick reaction in the morning on weekdays - the addressee rightly believes that he has all day ahead. The answer speed greatly depends on the age of the addressee: if adolescents respond to half the letters in 13 minutes from the date of sending, then 36-50-year-olds take 24 minutes.

At the same time, the most likely response time is only 2 minutes. And the most popular letter length is only 5 words.

The proportion of unprofitable letters depends on the load of a person - it seems to be clear and so, without calculations, but the numbers say that the manner of communication changes very sharply. He who receives more than 100 letters per day, he responds on average to every twentieth (5 percent). And those who do not put pressure on the psyche - for every fourth (25 percent). If you limit the sample only by letters that are the real people (and not spammers and not robots of different useful services), the result will be similar.

Conclusions could be reformatted in the list of useful tips. Do not write on weekends, late at night and in the morning. Specially focus on the time when your addressee is not at the computer (but walks with a smartphone in your pocket), if the answer is required fast and short. Avoid invested files when you want a quick response (half of the letters without investments respond faster than 32 minutes, and for letters with attachments this indicator is almost twice as much - 56 minutes). Finally, do not pull with a letter for more than a day: during the day they respond to 90 out of 100, and among the 10 remaining unprofitable interest, few are pleased.

Before the final letter in the chain, for some reason, people withstand a particularly long pause.

Long before all statistics, the Internet of the 90s tried to develop such a set of rules of decency, “Netiket”, network etiquette, where an email was given a special, honorable place. Netiket ordered to beginners the norms of behavior on the network - suggesting that everyone who has long been behaving on the Internet has already behaved. And only the analysis of the “big data” many years later revealed that the norm is actually.

But it turns out that even without strict universal rules, we generally cope. The brain on the machine takes into account the interlocutor’s presentation of the rules of decency in electronic correspondence. “Big data” show our ability to unconsciously adapt to the addressee. We begin to copy his pauses between letters and even his style. If the interlocutor answers quickly, then we try to keep the same pace - although this turns out, as a rule, only to the middle of the chain. If he rarely uses “I”, our answers become more impersonal. The same thing happens with auxiliary verbs, articles (analysis of English texts) and other style markers.

Most likely, this is true not only for electronic correspondence, but also for any form of communication between people - from conversations in the smoking room to meetings of the Boardes of Directors. It's just that this communication is much more difficult to analyze, since it practically does not leave traces. Moreover, in such quantities, so that you can make statistically reliable conclusions. With e -mail, everything is much simpler.

The authors nevertheless emphasize that even no one previously analyzed emails on such a scale. The concern for the secret of correspondence usually interferes. The only well -known exception is an array of 1.6 million letters related to the bankruptcy of Enron , which were placed in the public domain back in 2003. The US authorities found a scandal related to a fraudulent scheme, quite resonant for declassifying the internal correspondence of the company - and thereby made a real gift to scientists specializing in “big data”. Enron's letters are already devoted to several dozen scientific articles on communication theory . In particular, they helped create smart sorting algorithms like those that Gmail uses to decompose letters from folders or mark them as “important” ones.