While reading Geddy Lee’s My Effin’ Life, I was struck by his story of how Neil Peart listened to all of their albums while he was fighting cancer. It also reminded me of various rankings of Rush studio albums and how I agreed with some but differed with others. It was time to give the entire studio catalog another listen and come up with my ranking. That’s where the nerdery comes in.

Being a scientist, I felt that the album ratings must be based on something specific; they must follow a consistent protocol. The idea was that I would rate every song on each album by a consistent set of criteria, and the album’s rating would be the time-weighted average of the song rating. In other words, the songs wouldn’t count equally; a long, highly-rated song would count for more than a short, highly-rated song.

The question was what that “consistent set of criteria” for rating the songs would be. Bare 4–5-star or 1–10 systems hide what those scores represent. What makes a tune a three-star song versus a four-star? Some definitions are needed.

The endpoints of the scale were easy enough to figure out. There are songs I will nearly always skip over, and there are songs that I’d want with me if I were marooned on a desert island. But what about the middle? The idea of skipping or fast-forwarding stuck with me: some songs are great (although not desert-island material) that I rarely skip, and other songs are okay, but I am often tempted to skip or fast-forward through them. Those give me four steps: nearly certain to skip, good likelihood of skipping, unlikely to skip, and desert island. Four categories, not overthought.

It was surprisingly easy to use this scale, and I rarely felt that songs fell between the cracks or that I needed halfway points. Skippability was the crucial decider. So, off I went, listening to the entire studio catalog, although I didn’t listen to the albums in order.

Once I rated all the songs on each album, I calculated the album rating. This was done by multiplying each song’s length (in seconds) by its rating, adding all these values, and dividing this sum by the album’s length (in seconds). This produced a number on the 1–4 scale for each album, in other words, a time-weighted average.

Here are the results:


Moving Pictures is the highest ranked, as you would expect of the masterpiece: every song on it is on my desert-island list. 2112 and Permanent Waves follow, and no surprises there. Unexpectedly, Grace under Pressure comes next; I was surprised at how many songs on it I liked so much. Rush is also strong, which is a little surprising given that it’s their debut album and it predates Neil’s arrival.

At the bottom end, Caress of Steel brings up the rear, weighted down by the long Fountain of Lamneth. Bastille Day and Lakeside Park are the only tracks I like on it. It is followed closely by Test for Echo, which I’ve never liked: musically and lyrically, it has always seemed that the boys were mostly uninspired. Those two albums stand out; the next lowest ranked is Presto, but it is much better. It, Snakes & Arrows, and Counterparts are all a step down from the rest of Rush’s catalog.

Overall, most albums stand up well. The vertical dashed lines correspond to where an album would fall if all the songs were ranked 2 (often skipped), 3 (rarely skipped), and 4 (desert island). Most albums lie somewhere between 3 and 4, so ranking them is a bit of hair-splitting: they’re all good. Only six albums fall below that, with two approaching the mostly skippable mark (Caress of Steel and Test for Echo), and that matches my listening: I realized it had been years since I listened to either all the way through.

Two albums ranked differently than what I would have expected. I have good memories of Roll the Bones and Clockwork Angels, and as the rankings shook out, they pulled in lower than I would have guessed. On relistening to them, I realized my impressions were driven by a couple of songs I really like and that there are several songs on both that I tend to skip.

Desert-island songs

Of the 165 songs on Rush’s studio albums (not counting Feedback), a full third (55) make my desert-island list:

The skippables

Amazingly, I considered few songs (3.5%) immediately skippable. The last four of these are from Test for Echo.


All analyses were performed in R, and the data file and R source code are available. The last column of the data file contains all the song ratings; the code will work with any rating scale and similarly constructed data file.