Editor’s note
Lately, my younger two children have begun co-opting my phone, so as to text their older sibling, who is away at college. Their communications are mostly what you might expect, frantic gestures of affection and well-wishing that remind me of the way people used to wave handkerchiefs in the air, after ships departing from docks. Their text chains are filled with memes and GIFs, pictures of newborn puppies sleeping in piles, clips of young polar bears embracing each other and then falling over and tumbling down slopes. I miss you, they all say, or something to that effect. Come home. A few weeks ago I came across one exception—a series of curious messages sent by my ten-year-old. I will always and forever love and miss your beautiful smile, one message read. I will never be forgotten and you are the only one who can see my heart. I will always and forever cherish and always love you and love your life. I was disappointed, much in the same way other parents speak of being disappointed when they discover vapes in their children’s pockets. But to my mind, this was even worse—this was bad writing. Eventually my daughter explained to me that she had composed these messages using the predictive text feature on my phone. She typed in a word and the phone suggested the next word, based on an algorithm that aggregated all the language we sent to one another, and then crunched its numbers to determine the most likely pairings. She wrote I and the phone suggested will. She accepted will and the phone suggested always, and so on. Thank God, I thought. I will always and forever be grateful you didn’t write this. I suppose what I was objecting to was the defining feature of the algorithm, which functioned by offering up the most commonly-used