HOW RATINGS WORK:

Q:: WHAT ON EARTH IS YOUR RATING SYSTEM??!! WHY ARE THERE NUMBERS AND LETTERS??!!!

A:: The reason that both letters and numbers are used is that the rating system is in hexadecimal.

Q:: Hexawhoozit??

A:: Hexadecimal is the expression of numbers in base-16 (decimal is base-10). The base-10 integers 1-9 are expressed the same in hexadecimal. The base-10 integers 10-15 are expressed in hexadecimal using the letters A through F. The base-10 integer 16 is expressed as 10.

Q:: Ok ... so what do each of the numbers mean in terms of album quality?

A:: The breakdown is roughly as follows:

10 (16 base-10): The pantheon of my collection. Only about a dozen albums crack this tier. A++.
F (15 base-10): All-time great. Potentially in my top 15. No real reason for failing to crack the 10 barrier other than gut feeling. A+
E (14 base-10): Great, close to the all-time level. Potentially in my top 40. Solid A to low A+
D (13 base-10): Great, bordering on very good. Potentially in my top 100. Solid A- to low A.
C (12 base-10): Very good, bordering on great. Potentially in my top 175. Definite A-.
B (11 base-10): Very good. Potentially in my top 250. B+ or A- depending on my mood. Any album with this rating or above should be considered to have a "must own" recommendation from me.
A (10 base-10): Very good, bordering on good. Low B+.
9: Good. Solid B.
8: Good, bordering on mediocre. Solid B- to low B.
7: Mediocre, bordering on good. Solid C+ to low B-.
6: Mediocre. Solid C to low C+.
5: Mediocre, bordering on bad. Solid C- to low C.
4: Bad, bordering on mediocre. Solid D+ to low C-.
3: Bad. Low D- to low D+.
2: FAIL. F.
1: BEYOND FAIL.
0: I don't even want to try to comprehend what kind of album would get a 0 on this scale.

Q:: Wait a second. Shouldn't an 8/16 correspond to a 5/10, which would be a mediocre grade? Why are you giving "Good, bordering on mediocre" a 5/10?

A:: You should not think of an 8/16 as a 5/10. The biggest thing to keep in mind with this scale is that the difference in quality between successive grades is not consistent throughout the scale, and this is intentional. The difference between a 10 and a C is about the same as the difference between a C and an A, and about the same as the difference between an 8 and a 7.

Q:: Why would you have an asymmetric scale? Why not a 1-10 scale with 5 as mediocre, 1 as awful and 10 as great?

A:: If I chose bands completely at random, such a scale would make sense. It is most likely that the distribution of my ratings in such a circumstance would approximately form a normal distribution, with a 5/10 mediocre grade at the center. The reality, though, is that I review bands from my collection, and my collection naturally skews towards groups who make albums that I like. The result is that such a distribution would have a heavy concentration of 7/10 and 8/10 grades, with a long tail providing a large negative skew. Setting up the distribution up in the way I have done largely corrects this problem.

Mathematics aside, the system I've set up has another advantage in relation to the nature of albums I review. One of the main reasons to have a rating system at all is to establish approximate "tiers" in my tastes. These tiers are not ironclad (it's possible for me to enjoy a B-rated album more than a C-rated album sometimes) but they provide a good breakdown of how my tastes line up most of the time. When setting up tiers, it's far more interesting to have increased precision in rating good albums than in rating bad ones. Hence, while the first 8 grades (1-8) span the equivalent of a 0/10 to a low 7/10 (I don't need to have a lot of precision in determining just how mediocre or bad an album is), the remaining grades span the equivalent of a solid 7/10 to a high 10/10.

Q:: Didn't you used to have a rating scale with two numbers, one on a 1-10 scale and one on a 1-15 scale?

A:: Yes I did. I first developed this combined scale in early 2000 and stuck with it for almost a dozen years. It was an attempt to synthesize the scales of Mark Prindle and George Starostin into a system that took advantage of the best features of each, and I'm rather proud of how well it worked. You can still see several reader comments that make reference to grades in that system, and I will not be modifying those existing comments.

Q:: If you liked the old rating system, why would you completely ditch it?

A:: There are a few reasons I decided to change the system. One is that change, every so often, is fun. Another is that, as long as I kept the 1-15 system, I would have trouble ditching the "George clone" label that has plagued me since the beginning (it's not as deserving now as it once was). Another is the sheer novelty of having a scale based in hexadecimal. The biggest reason, though, is that I could never figure out a good solution to one irritating feature of the 1-15 scale: the lowest grades (1,2,3) were almost never used. Indeed, even the worst rated album in the current system (receiving a 1) still managed a 3/15 on that scale; it was nearly impossible to conceive of an album that could get a 2 or a 1. Now all of the ratings in the scale (apart from 0, for now) are represented.

Q:: I fear change. If it were up to me we'd all still be sitting on the floors of caves, trying to decide if rocks were edible. Any chance you'll change the rating system back?

A:: Nope, no chance.

Q:: Can you at least post a guide to how the new ratings correspond to the old one?

A:: For each grade, I will provide the corresponding Starostin-style and Prindle-style rating below:

10: Starostin: 15. Prindle: 9 or 10.

F: Starostin: 15. Prindle: 9 or 10.

E: Starostin: 14. Prindle: 9 or 10.

D: Starostin: 13. Prindle: 9 or 10.

C: Starostin: 12. Prindle: 9 or 10. These are essentially a coin-flip between a Prindle 8 and a Prindle 9, landing on the 9 side.

B: Starostin: 12. Prindle: 8.

A: Starostin: 11. Prindle: 8.

9: Starostin: 11. Prindle: 7.

8: Starostin: 10. Prindle: 7.

7: Starostin: 9. Prindle: 6.

6: Starostin: 8. Prindle: 5.

5: Starostin: 7. Prindle: 4.

4: Starostin: 6. Prindle: 3.

3: Starostin: 5. Prindle: 2.

2: Starostin: 4. Prindle: 1.

1: Starostin: 3. Prindle: 0.

Q:: How many albums have received each rating?

A:: The breakdown is as follows:

Total: 1093

10: 11

F: 9

E: 37

D: 110

C: 89

B: 120

A: 148

9: 144

8: 166

7: 91

6: 83

5: 40

4: 25

3: 15

2: 4

1: 1

Q:: Could this get any geekier?

A:: Yes it could! Here are the distribution statistics (base-10) for my album ratings:

Mean: 9.3687

Standard Deviation: 2.7341

Skewness: -0.0959

Kurtosis: 2.6600

Standard Error: 0.0827

Back to the main page