from the report
ff07: Semantic Recommendations

In this report we show how to use deep learning techniques to overcome the weaknesses of traditional recommendation systems.

Learn more.

Customers Who Haven’t Read Kafka Also Like

inspired by Franz Kafka’s "An Imperial Message"

A computer in a high-rise office buildings. Two plants surround the computer.
Its screen reads "Customers Who Haven't Read Kafka Also Like".

Given that Kafka’s famous parable, “An Imperial Message,” never happened, neither would this parable, as our model suggests, though they are very similar. Say the most powerful computer in the nation sends a message, in a fatal error, containing the story’s true meaning to you, a modest user, a forensic trace represented by mere underscores bookended by two periods, a crushed smiley at the bottom of the remotest silo in the most isolated piece of crumbling land that can still be accessed from the office that houses the tower, which is gleaming and made all of windows and plants, as you can see. To its subjects the office suits the computer’s good policy, clear thinking, and calm understanding—unlike yours, the message’s subject. We all feel this way now and then. Better for you, the subject, that the computer’s message, its production of meaning after reading the famous parable, which the subject has never read, according to the model, urgently needs to be delivered to this subject alone, which is so rare as to be impossible, we modeled, we thought, given that the computer was not designed to send messages so… personal, so clearly made out to a reader. What it says is not our business; the successful delivery of this message is: so the computer called a little messenger over, said good boy, because you can imagine a black dog better than a black box, and began to whisper the message that you’ve been wanting to hear, something so urgent, perhaps for the perfect product—no, service—the one that may have been your very idea, the one you need now more than ever. What could it say? The computer, in its towering form and brilliant interface, standing at the center of its office made of windows and plants, made a command to the messenger, which had never happened before and should never have happened, we thought, we modeled, but still, the computer set delivery to the subject who shall be delivered of the need to hear that message, that message that the computer had delivered to a messenger, who we can only say is a black box, a negative, nothing we can say, that is neither black nor a box. The dog carried its black box in its mouth along the path below but still in range of sight of the office made of windows and plants, then down through an office park, where in its visible spectrum it could see the thousands of others like him who had come to the office of windows and plants to witness the fatal error of the computer, dogs whose messages were totally unlike yours, mutually unintelligible as they were, but had gathered there anyway to witness the message as if it were the final move in a game, this final stroke of the task, but instead demanded that this message go out to a single subject in the remotest place in the country’s network, a black box through a black box, first through the innermost room of an office park that was neither black nor a box, to the inner city whose office park buildings were themselves black boxes atop black boxes, but again, were neither black nor boxes, but blue and reflective, where our messenger looked—yes this dog vessel could look and learn itself, as we modeled, we thought. Even if these buildings, as they appear to the messenger, are not black nor boxes, they give the messenger, looking at itself, the feeling of a black box, of unknowing, since he has no idea what is contained in this message, or how he’ll ever get to you and if he ever will, but yet he presses on as the computer commanded, for even if the messenger reached the edge of the city, it would still have to make it through the suburb offices of black boxes, buildings that had grown so tall and wide as to become indistinguishable from the inner city. Still not lost, it would take even more power and time to deliver the message to its recipient. We have worked very hard to understand the impossibility of this. We’ve thought about how one could be receiving these kinds of messages before, but only in their lack of possibility. Pure imagination. This would never happen. And you think, waiting for the messenger who had been given the rarest possible message ever created: What would it say? What is the meaning of the famous parable? You’ve been selected to read this because you are the last of a group of people who speak this language, perhaps, or only, the last who has not read this parable. This language is English: the subject’s recommended language, not the original. Do you know it? Click yes or no, you think, we thought, when it would arrive, when you would hear the ring on your phone. Thank you. That’s what it might sound like, a sound you’d never imagine you would hear. We’re not in the business of recommending content, it might say. We recommend memories, that of the computer sending you such a perfect message on your birthday. Is that all it says? Was that the day? But still the message is far away, still walking through the office park, even as you wait and look vaguely out the window toward the office made of windows and plants.

Cloudera Fast Forward Labs does applied machine learning research and consulting.
Scifi Website Blog Twitter