Tuesday, January 21, 2020

The algorithm can’t predict the future!


Well the tech guys want us to believe that the algorithm is that magic that it actually can predict the future in the big data. Maybe in a George Orwell future, it can, but I really don’t want to live there.

I got this idea for this text from a press release of another company doing site with an algorithm that goes on the mood. In this case for sync placements. For me, that feels like 2010 when several companies did that. If you really want that search engine goes top Spotify that found out that people that want playlists just want mood playlists. And yes, sad playlists are the most popular right after workout playlists. The problem that now occurring on Spotify is that the audience is not6 listening to the music they just use it as background music on other things. It’s like the music in the ceiling of the supermarket. It’s there but no one really don’t care what it is you are too busy shopping.

What an advertising firm really want is not just mood music, it’s a cool artist with it. And if they can find someone cool as Taylor Swift for no money at all they will take it. But how can an algorithm find a cool artist? And especially cool artist when you have a low budget?

It’s all about the future. Let say I have an unknown Canadian band with a really nice song. But I also know that this band is booked for these three cool festivals. Also, I know they are recording for a new album that is released around the same time as the commercial and on top of that, I know they will be in the sofa of the best morning show. This info is really crucial. But the algorithm will only find out this (if ever) when it already has happened. This is not written in any public records. And when it has happened it’s dead on arrival. Yes, knowledge is power.

So, the only way an algorithm would know if its ear dropped on you. Or for speaking out loud, spying on you. Yes, we have all that experience that you talked to a friend about new speakers for the living room and boom hours later you get online commercials just around speakers. We know that big companies listen to us with their speakers, on our phones, etc. Even though none of them would admit it.

But that is the only way that an algorithm would have a chance to predict any future. Because if it just gets to the info when you search for these speakers online. The chance is just that you also bought them and the commercial you get later on is just annoying since you just bought your new speakers.
This only goes for really big companies like Google, Amazon, Microsoft. It won’t work when you need special people like supervisors. That is too narrow to get into. Spotify is to small to eardrop on you. And how on earth would they find out when you really start to like a song or a new artist? If that is going to happen the algorithm has to be implemented in your brain. And I hope we have sense enough to stay away from that.

The algorithms today mainly just goes on old info. The algorithms would say that this artist is totally out, but don’t know that the artist might just wrote the song of their lives. In that way recommended the whole wrong thing to the buyer. They would have loved to be part of that great song that just was written and have their brand connected to it. The artist might have liked the brand and could get some extra marketing money for it. I would have seen the algorithm that could predict that Queen was taking over Live Aid 1985. They were counted out as going away. But their live experience took over and they came out as a new hot band.

I hope we never goes so far that the algorithm should be able to predict the future. It’s enough that it spies on what we do and learn from that. That is still a guess, to leap that it can read us is not a pleasant thought.



No comments:

Post a Comment