I've always preferred camel caps to underscores when it comes to naming conventions, but I've never really known why. Recently I've been writing a lot of Python, where the standard is to use underscores for most names, and now I realise why I don't like underscores. I hope to make my argument for camel caps objective — or, failing that, at least better than the mouth-frothing, religious reactions you normally get in these “X is better than Y” type of discussions.

Line Width

If you are adhering to a line length limit, then underscores will encourage less descriptive names, and more temporary variables.

The Style Guide for Python (PEP 8) says each line should be no longer than 79 characters. Style guides for other languages often have similar limits on line width.

When your function names and variables are descriptive, you end up hitting the maximum line width pretty often. Underscores make the situation worse. I find myself shortening names just to fit the line width, and that is not a good thing. You can break long statements into multiple smaller statements, but then you have to introduce temporary, single-use variables. Adding unnecessary variables isn't ideal, either.

Python is especially bad in this respect because whitespace is meaningful, unlike most other languages. If you want to wrap a long statement onto multiple lines, it looks very similar to a loop or conditional structure. On top of this, sometimes you need to use backslashes to break up a line and other times you don't.

You can see all of these problems in in a typical SQLAlchemy query:

dirty_col = db.species.c.num_dirty_occurrences
        num_dirty_occurrences=(dirty_col + newly_dirty),
    ).where(db.species.c.id == row['id'])\

Underscores Make Spaces And Other Symbols Harder To See

Compare this code written with underscores:

i_am_the = very_model_of(a_modern_major)
general_ive.information_vegetable = animal_and_mineral
i_know_the(kings_of, england_and, i_quote, the_fights, historical)

To this identical code, translated into camel caps:

iAmThe = veryModelOf(aModernMajor)
generalIve.informationVegetable = animalAndMineral
iKnowThe(kingsOf, englandAnd, iQuote, theFights, historical)

I would argue that the camel caps code is more readable. The underscores look too similar to spaces, which makes the spaces harder to see. Camel caps make the other symbols stand out, which makes the code easier to scan with your eyes.

I know a programmer who insists that underscores are easier to read than camel caps because humans are used to reading words separated by spaces. My problem with this argument is that code isn't written like a novel. When you scan code with your eyes, you are primarily looking for syntactic structure, not names. You want to instantly see where a loop starts and ends, or where a functions arguments start and end, et cetera. Only after you've mentally digested the structure do you actually care about the names of variables and function.

For example, let's say you are scanning a file to find a certain method. First you scan for the class you want. You ignore every name in the code except for class names. Once you've found the class, you start scanning for method declarations. You ignore every name in the code except for method names. If the class names and method names blend in with all the other code, that makes for a frustrating experience.

What Advantages Do Underscores Give You?

Names with underscores are probably easier to read than names in camel caps. As I mentioned above, I don't think you should be optimising for the readability of names, at the expense of the readability of the syntactic structure of the code.

If you are using a language where underscores are a widely accepted standard, then you probably want to stick with that standard. Having a single naming convention is better than having multiple naming conventions, even if the single convention is suboptimal. However, this doesn't mean that underscores are inherently good, it just means that you should pick the lesser of two evils.

So, what advantages do you get from using underscores instead of camel caps. I can't think of any others.


In the end, the camel caps versus underscores issue is fairly minor. After working in a code base with multiple different conventions, I've realised that doesn't matter a whole lot. Sure, the underscores are slightly more annoying to dig through, but it's not like it brings development to a grinding halt.

If you are starting with a clean slate and have the opportunity to choose, then choose camel caps. If not, there is no need to make a big deal out of it.