‘Rearview thinking’ has been described as the way humans sometimes try to grasp the road ahead by looking through the rearview mirror rather than the windscreen at the actual road in front of them. Put another way, we are always filtering our understanding of the new, through what we expect from the old. Add statistical analysis and you get Polyphonic HMI ( www.polyphonichmi.com ), a company dedicated to the science of hit predictions, who claim they can accurately recommend which music will appeal to an individual or entire market. Record companies and artists are already hiring them to improve their chances of a hit. CEO Mike McCready explains your taste in music:
>Where does your music database come from, and your charts analysis?
Primarily it comes from retailers we work with since we’re now going to be launching our Human Music Interface, our own special version of music recommendation technology. It also comes from radio stations we work with, and from companies who specialize in having digital databases of music.
>How is HSS set-up to deal with the ongoing and inevitable shift from to giant, widespread popularity, to more specialised niche tastes?
We can profile the taste of any audience. With HSS in most cases we’re profiling the tastes of a mass mainstream audience but we can profile the tastes of listeners of a single radio station or even a single individual. We even have a tool called a Music Taste Test that can do exactly that, profile a single user’s music taste.
>How much Australian hip-hop do you have in your collection, and how well do you think you are placed to predict which Australian hip-hop songs will prove to be ‘hits’ ?
The only Australian music we have in our databases would be those songs that have had some international success at this point, or at least been released in the US or UK. As soon as we have a couple of label clients in Australia we’ll go ahead and take the plunge into that market. It will be really exciting because Australia has historically produced such great artists and music.
>2 tracks that’ve surprised you by getting low HSS scores?
Lyrics and artist image are two big determining factors we don’t touch. That’s still up to the A&R and marketing guys. So some songs that have been very lyrically driven such as Eminem’s Cleaning Out My Closet became a big hit without scoring well on our system. Nevertheless, we know how to factor for that sort of thing and by working with the labels we can help them make good decisions.
>2 tracks that’ve surprised you by getting high HSS scores?
Our technology isn’t even really that good at telling you if a song “sounds” like a hit. What it IS good at is telling you if a song that people already thinks “sounds” like a hit will perform in the market.
>Given your prediction technology is based on analysing the songs of the past, how can it be expected to cope with *new* musical genres that may spawn hits?
Since 90% of hits over the past 50 years have conformed to the same large but limited number of mathematical patterns and in that time many new genres have emerged (pop, grunge, disco etc) it can be said that if a new genre of music were to be invented tomorrow, most likely HSS would be able to determine which singles from that new genre would chart and which ones would fail.
>Do you find many differences between what HSS defines as a hit and what you actually prefer to listen to at home?
Hit songs are my specialty and that of Polyphonic so I spend a lot of time with them. However, yes, from time to time you have to think about something other than work and there is a LOT of quality music out there that will never be hit singles. I have a huge collection of jazz getting a lot of rotation at home.
>HSS studio optimisation? How do you liase with artists to ‘improve’ the score of their potential ‘hits’?
Most songs destined to become singles come out of the studio already sounding great. However more often than not the scores are just a little bit off and that can prevent the song from performing in the market. We can suggest very minor tweaks in the mix and production values and usually correct the problem. We don’t get involved in the composition and arrangement. We leave creativity to the artists and are ONLY concerned with giving that song the little extra polish that will propel it up the charts.
>What do you think about the recent Australian invention which predicts which type of beer you will like best, by taking a mouth swab at the pub entrance?
Interesting. I can see uses for that, especially if I already don’t know what kind of beer I’m going to drink in a pub. If they get it right I’d be happy. If they get it wrong I’d be upset that I didn’t order something I know I’ll like. I’m not likely to get tired of my favorite beer as quickly as I am a hit song. So, my need for trying new beers is lesser than my need to find good new music on a regular basis.
jeanpoole