What is 99spokes.com algo for spec level score? Do you trust it?

I love 99spokes.com but wish they could improve a bit.

99spokes gives "spec level" scores to most recent bikes and ebikes and plots them on a scatter plot. Just go to any random bike page and look for this plot. The price history can be telling too albeit limited in dates. The plot is useful if a) you could trust their scores fully and b) they didnt have any errors (e.g. they mix up the prices for diff polygons). The scatter plot tells you that if you like a certain bike, there may be other ones with better spec scores at the same price.

But aside from tiny errors and trustworthiness of the spec level, another peoblem is that the plot is using msrp prices and not the current sales prices. The scatter would look very different since some companies like specialized are having deeper discounts than others like polygon who were economical to begin with.

But what is their algo for rating so many bikes? Hand scores or coding?

Now, one rough and lazy method for scoring all components in existence on the market is to use price as a measure of quality. Feed numerous pages of component prices into a code and normalize by a max price x10. Say a drivetrain is worth $300 retail while the most expensive one is $1000? No problem, the expensive one gets a 10 and $300 gets 3.

If you click on the spec level, you get a dropdown breakdown of the scores for frame, drivetrain etc.

So … what is their algo? And do you trust it plus minus say x% error marging?

Here is a ridic comparison of a lot of bikes and ebikes that shd not be compared, but there for you to go deeper in any you like.