Is band name scarcity a myth? A quantitative approach, revised and expanded

While there has been little empirical study of the phenomenon, it is a widely held belief among aspiring musicians that ‘all the good band names are taken’ (multiple anonymous informants, pers. comms). Given that there is also a belief that a good band name is important (Indie Professor, 2011), there is significant ‘band name anxiety’ among young musician. Whitman (n.d.) has already demonstrated that a constrained, ‘grammar average’ randomly generated list of band names can produce viable unclaimed options. However, Whitman does not provide more than a wild guess (99%) as to the probability that a randomly generated word string will be a viable unclaimed band name.

It is therefore worth examining whether band name scarcity is an actual phenomenon. Here we show that band name scarcity is largely an imaginary phenomenon and that the number of random word strings that are actual in-use band names decreases rapidly and non-linearly as the number of words in the string increases. We conduct Google ^TM searches of randomly generated word strings of various lengths and find that while up to 70% of randomly selected one word strings are already a band name (sometimes attached to more than one band), 0% of randomly selected two and three word strings are band names. In addition, 0% of randomly selected three word strings return any Google ™ hits at all. Despite this, a reasonable number of random two and three word strings would actually be pretty awesome band names and definitely seem to be up for grabs.

Methods

We used the first random word generator returned by Google ^TM  when searching for ‘random word generator’ (https://randomwordgenerator.com/). We generated 20 random strings of one word, 20 strings of 2 words, and 20 strings of 3 words (see appendix I for full experimental data). We then searched Google ^TM to determine whether the strings corresponded to existing band names.

In the case of one word strings, we initially searched for (word)+band. We did not count cover bands in the results, as this would have severely reduced the prestige of our study. We only accepted exact matches and did not include bands for which there was, for example, an article in front of the randomly selected word. For instance, a band called The Word did not count as a match for Word (not (word) as in the abstract placeholder for words in our study, but the actual word Word), which happened to be randomly selected as the search word for trial 13. But of course it follows that we did not accept (article)+(word) in general, anymore than we accepted The Word). Our method was case-insensitive, so hAND was accepted for Hand (trial 17), for example.

In the case of multi-word strings, we did not accept slight variations of word order, although it was extremely tempting to do so, as in the case of the 57^th trial (Vertical Nightmare Digital). Neither did we give in to the temptation of partial matches between one word and multi-word strings, such as in the case of Torture (trial 15) and the Dutch band Severe Torture, which we didn’t count, even though we wanted to. We could have come up with a system of points for partial matches, but this would have produced an unacceptable number of degrees of freedom in the analysis and reduced our overall confidence in our result as well as their power.

Methodological challenges and limitations

Through experimentation, we determined with a high degree of certainty that the random word generator does not produce Proper nouns, but only Common ones. Names of persons and places, except those that refer largely to natural features (e.g. Underhill, Riverside, etc) were therefore excluded from the study and we cannot determine whether that particular semantic field suffers from band name scarcity. For the same reason, we cannot determine whether the names of famous people and places that would make great band names are all taken. Our study deals primarily with common nouns, but we figure that’s already a pretty good start.

It became clear early on, especially in the multi-word strings, that our initial plan of searching for (word)+band or (string)+band was inadequate, and we had to revise our methods during data collection in the light of empirically determined conditions. I’m not saying we made it up as we went along, but pretty close. For example, even two-word strings such as (Option Campaign)+Band produced annoying numbers of results completely unrelated to bands, but we did learn a great deal about CISCO’s Outbound Option Campaign Manager. The string Layer Glacier proved equally difficult to constrain, as apparently, there are layers in glaciers, and geologists seem to think this is worth publishing endless numbers of papers about.

Three word strings proved even more problematic. They still had nothing to do with bands, and none contained the exact search string. We elected to constrain the search by enclosing the string in double quotes to force Google ^TM to find an exact match, and appended the word band outside the double quotes. This alleviated some of the problem of weeding through thousands of non-band related results, but did produce some much more specific and sometimes unfortunate outcomes. For example, trial 39 (“Breeze Medicine”)+Band had to be aborted with the Task Manager to avoid a ransomware attack, sadly interrupting at the same time the very groovy Eloy album we were listening to.

Given the absence of any results at all for string lengths greater than one, we belatedly realized that instead of conducting a set number of trials for each string length, we should have just searched until we got a hit at each string length, and used the number of trials required to get a hit as our main metric (with an arbitrarily defined maximum number of attempts, obviously). But by then it was almost dinner time, and given the way things were going, it would have taken all night. Automation seems an obvious solution here, but we decided to let younger and nimbler minds, in future studies, build on our pioneering, if preliminary, work.

Results

We only found actual band names when the string length was equal to one. None of the multi-word strings were actual band names. None of the three word strings actually produced any Google^TM  hits at all. Which is pretty lame, but an interesting result in itself.

Figure 1: None of the multi-word strings were actual band names

Discussion

Given these results, and given that several of the multi-word strings would actually make pretty sweet band names if anyone used them, we can confidently say that for a large part of the potentially infinite parameter space for band names (Shultz 2014) there is little, if any scarcity of band names, and there are still lots of way cool choices. While the proportion of one word strings already taken as band names is high, there are still some viable options among the unclaimed words, such as Conceive, or even Philosophy (which we didn’t count as a positive result, as it was only a cover-band). Incredibly, Plagiarize was not taken, despite its post-ironic potential and its natural resonance. For some reason we can’t fathom, Word was not taken either, although there were numerous variations among the bands we found (notably, The Word). Still, we successfully resisted the siren song of data fudging, if ever so distantly heard.

One of the surprising and counter-intuitive results of our study is that so many cool potential band names were generated in so few random trials. For example, while Whisper Veil produced many music and art industry related near misses, no band seems to have adopted it. We had high hopes for trial 30, Grant X-Ray, but while many astro-physics papers came up, no astro-physicist seems to have formed a band of this name. Shrink Hole, being related to metallurgy and industrialization, likewise caused great anticipation as a potential Metal band name, but produced no results.

The three word strings produced even better and more numerous viable band names. As an aside, even though we had no results to two word strings, we did the threes anyway, just in case the curve was U-shaped. It wasn’t.  But the potential names alone were worth it, even though they got no hits at all. Not just no band related hits. No hits at all.

The fact that Incredible Ideology Sale and Pig Breakdown Tower are up for grabs should be enough to dispel any notion that band name scarcity is a real phenomenon. How incredibly cool are those two? It should also immediately cause us to wonder about the quality of the human imagination. And we won’t even mention Annual Polite Symbol, Projection Pyramid Experience, and Far Hope Attachment. Perhaps less surprisingly, even one of the two word strings (Dump Mislead) produced only a single hit, while Lesson Photocopy led us only to a phytoplankton microscopy lab assignment. Finally Threshold Bat Supply, while not quite as marketable as the previous examples, is only one consonant away from meaningful viability. In fact, in a future paper, we plan to test whether the parameter space is characterized by a correlation between the number of words in the string and the number of viable band names.

As a preliminary test, we assigned a Bayesian Band Name Viability Score (BBNVS) to the randomly generated pseudo-band names, we determined that one word strings have uniformly high band name viability (Avg = 1.00, StDev = 0.00), and that three word strings (Avg = 0.69, StDev = 0.36) have higher average band name viability than two word strings (Avg = 0.55, StDev = 0.32). Of course, these are only our prior estimates of BBNVS, and we would have to develop a metric that has better interobserver subjectivity before we can get a more meaningfully study the relationship between string length and band name viability.

Conclusion

We have shown that band name scarcity is largely a product of the imagination of aspiring musicians. Good one word names may be hard to come by, but there are still kick-ass multi-word options out there.

References

Indie Professor, 2011. https://www.theguardian.com/music/musicblog/2011/mar/22/sxsw-bad-band-names

Schultz 2014. https://www.researchgate.net/post/Are_there_infinitely_many_possible_sentences_in_a_natural_language

Whitman n.d. http://alumni.media.mit.edu/~bwhitman/10000.html

Appendix 1: Experimental data

Trial	NumWords	Words	Band	Year	Country	BBNVS
1	1	Conceive	0			1.00
2	1	Apple	1	1968	Wales	1.00
3	1	Neighbour	1	2013	Australia	1.00
4	1	Relaxation	0			1.00
5	1	Agent	1	2008	New Zealand	1.00
6	1	Meet	0			1.00
7	1	Philosophy	2	2010	England	1.00
8	1	Ghost	1	2008	Sweden	1.00
9	1	Plagiarize	0			1.00
10	1	Graduate	1	1978	England	1.00
11	1	Rhetoric	1	2013	USA	1.00
12	1	Remark	1	2009	Russia	1.00
13	1	Word	0			1.00
14	1	Extent	1	2010	USA	1.00
15	1	Torture	1	1986	USA	1.00
16	1	Zero	1	1998	India	1.00
17	1	hAND	1	2004	England	1.00
18	1	Clay	1	2014	England	1.00
19	1	Utter	1	2010	Portugal	1.00
20	1	Pump	1	1979	England	1.00
21	2	Option Campaign	0			1.00
22	2	Layer Glacier	0			0.25
23	2	Image Nuance	0			0.30
24	2	Warm Incident	0			1.00
25	2	Legislation Fame	0			0.80
26	2	Wait Achieve	0			0.10
27	2	Lesson Photocopy	0			0.20
28	2	Upset Dip	0			0.10
29	2	Guide Credit	0			0.30
30	2	Grant X-Ray	0			1.00
31	2	Prince Get	0			0.10
32	2	Ruin Document	0			0.50
33	2	Dump Mislead	0			0.30
34	2	Shrink Hole	0			1.00
35	2	Player Cattle	0			0.30
36	2	Variation Prejudice	0			0.60
37	2	Wheel Thread	0			1.00
38	2	Development Punish	0			0.05
39	2	Breeze Medicine	0			0.75
40	2	Whisper Veil	0			1.00
41	3	Certain Stable Dribble	0			1.00
42	3	Brink Know Serve	0			0.30
43	3	War Fairy Speed	0			0.60
44	3	Incredible Ideology Sale	0			1.00
45	3	Pig Breakdown Tower	0			1.00
46	3	Wilderness Finish Soar	0			0.30
47	3	Inquiry Tough Troop	0			0.50
48	3	Threshold Bat Supply	0			0.40
49	3	Projection Profession Nest	0			0.25
50	3	Exclusive Agile Lift	0			1.00
51	3	Annual Polite Symbol	0			1.00
52	3	Cherry Base Ambiguous	0			0.85
53	3	Simplicity Review Abbey	0			0.50
54	3	Demand Dinner Absolute	0			0.75
55	3	Projection Pyramid Experience	0			1.00
56	3	Soldier Problem Buffet	0			1.00
57	3	Vertical Nightmare Digital	0			0.90
58	3	Freight Treasurer Trade	0			0.30
59	3	Far Hope Attachment	0			1.00
60	3	Chest Factory Summer	0			0.10