A lot of data problems aren't complex so much as they're just messy. For example, I recently wanted to make a list of bands with state names in their names. Or, more accurately, I wanted to look at such a list, and I knew how to assemble it the annoying way by doing 50 searches and 50x? copy-and-pastes, and the amount I didn't want to do that work vastly exceeded the amount I wanted to see the results.  

But I have tools. This is not at all an example I had in mind when I was designing DACTAL or integrating it into Curio, but it's the kind of question that tends to occur to me, so it's not terribly surprising that the query language I wrote handles it in a way that I like.  

A comma-separated list of state names is just data. (It's a little more complicated if any of the names in the list is also the name of a data type, but only a little.)  

DACTAL adapters move API access into the query language, so .matching artists traverses Spotify API /search calls as if they were data properties.  

Multi-argument filter operations are logical ORs, so :@<=10, popularity>0 means to pick any artist that is in the top 10 regardless of popularity, or has a popularity greater than 0 regardless of rank.  

Any time you try to ask an actual data question, as opposed to a syntax demonstration, you quickly discover that a lot of the answers are right but wrong. If you ask for bands with state-names in their names, you get the University of Alabama Marching Band, which is exactly what I asked for and not at all what I meant. So the long ~Whatever list in the middle of the query drops a lot of things like this by (inverted) substring matching. "Original" and "Cast" are not individually disqualifying, but they are when they occur together. "Players" generally indicates background crud, but not Ohio Players.  

And when there are multiple bands with the same name, /name.(.of:@1,followers>=1000) groups them and picks the most popular no matter how small that "most", plus any others with at least 1000 followers.  

So here's the query and the results. They're still messy, but they're close enough to the mess I wanted.  

 

If you just want the short version, we can get one artist per state out of the previous data like this:  

 

Why the two ..s instead of just .? If you know your state-name bands, you might already have guessed.  

Alabama
Alaska Y Dinarama
Arizona Zervas
Black Oak Arkansas
The California Honeydrops
Alexandra Colorado
Newfound Interest in Connecticut
Johnny Delaware
Florida Georgia Line
Florida Georgia Line
Engenheiros Do Hawaii
Archbishop Benson Idahosa
Illinois Jacquet
Indiana
IOWA
Kansas
The Kentucky Headhunters
Louisiana's LeRoux
Mack Maine
Maryland
Massachusetts Storm Sounds
Michigander
Minnesota
North Mississippi Allstars
African Missouri
French Montana
Nebraska 66
Nevada
New Hampshire Notables
New Jersey Kings
New Mexico
New York Dolls
The North Carolina Ramblers
Shallow North Dakota
Ohio Players
The Oklahoma Kid
Oregon Black
Fred Waring & The Pennsylvanians
Lucky Ron & the Rhode Island Reds
The South Carolina Broadcasters
The North & South Dakotas
The Tennessee Two
Texas
Utah Saints
Vermont (BR)
Virginia To Vegas
Grover Washington, Jr.
The Pride of West Virginia
Wisconsin Space Program
Wyomings