But we can. What we can't make is an objective measurement of whether something tastes good - much like we can't make a similar objective measurement for whether an image looks good, or a music piece sounds good.
The question in the article was about whether the taste of ketchup has changed over time. Now, can we measure the taste of something, record it, and compare it meaningfully to a new measurement? If there is a difference between both measurements, can we say if that difference is significant or not (i.e. would a human be able to tell the difference)?