There is a new pizza restaurant in my town. Not that we needed one: we had at least five already, but now we have a sixth. This one is less than a half mile from two other pizza locations (and arguably the two best), and I do wonder what the owners are thinking. They must think that they have something truly excellent if they plan on pulling that off, especially in New Jersey, a state where everyone has a strong opinion about what makes a good pie.
Honestly, while we might not have needed one, we could certainly have better places than we do. Two of the shops are really poor quality. Of the two good ones, the first is only OK, and the second is really tasty, but it isn’t terribly good pizza. (If you need a further point of reference, this means that I go to the one with OK pizza and wish it tasted better), and these are my consistent experiences over the past several years.
It seems that the new place, unfortunately, has managed to gain a negative reputation immediately. It has been open for less than a month and already it has a number of 1-star markings on Yelp. This means that they are going to be in the whole for quite some time before they can manage to get good reviews again. Unfortunately, they have no choice but to try to survive tyranny of the clock.
It is not uncommon for a restaurant to have a change of chefs, or a change of management which leads to better or worse food conditions. There is a Thai restaurant in the next town over which has some excellent dinner options, but that is only true because about a year and a half ago they got rid of their chef and replaced him with someone who could actually cook.
One of the projects I am working on is an attempt to let marketers reflect on the fact that recent data points really should be counted as superior to less recent ones. If there are two reviews on a product, and one of them was written five years ago and the other was written yesterday it is almost insane to represent the two values as being equal worth. All things being equal, the one five years ago is almost meaningless while the one written yesterday is incredibly relevant.
The problem actually reminds me of how polydimensionality can completely rendered in only one dimension. While that is often a useful tool, the problem is that the more data you cram into smaller dimensions, the most disparate your data actually becomes. Points diverge which would otherwise be touching in two or three dimensional space.
Of course, none of this should be news to anyone. Unexpected information is often discovered by looking at the holes in the lines, but I can’t help but think that we have found ourselves believing that our one-to-five ratings are as reliable as we claim them to be.