Here is an example I am working on: 
The left one if is visualy more balanced to me (it's compensating the off balance of the mark which is lighter on the right) and the right one is perfectly centered within the square.
If it wasn't a geometric-based thing I would go for the visual balance but in this case I am afraid someone would notice the off-center thing and it would annoy people.
What would you do?
Optical corrections are a feature of high quality graphic design. What is important is how things are perceived (typically to maintain coherence/alignments, and fulfill the viewers' expectations)
The positioning of elements in software is often based on their bounding boxes, which don't account for their more specific shape. Visually, a triangle will be much heavier at its base than at its apex, so when one is dealing with a "play" button (to tie in with the YouTube example below), centering the shape based on its (rectangular) bounding box would not distribute its visual weight equally.
YouTube's logo uses such an optical correction where the apex is pushed a bit further to the right to balance the visual weight more evenly across the center (more obvious in the left image)
As per @joojaa's comment below, it is possible to calculate the center of a triangle (centroid) and other shapes, but that's not typically something you'll learn in graphic design curriculum and you will eventually face limitations as the shapes become more complex.
In your design, I see the difference when I cover the other design from my sight and I think the current correction is a bit overdone as it stands. I would decrease it by half and reassess.
Also note that your current design may require further optical corrections. For example, shapes that are intended to look like a square may need to be made a bit less tall than they are wide to look like a proper square, compensating for the vertical-horizontal illusion.
Another relevant correction in logos is adjusting for the perceived weight of the icon and type when creating the black and white versions (positive and negative). The negative (white on black background) will look thicker than the positive (black on white background), so you would need to adjust one of the versions with a (very) light stroke (did I say very?) so they will look equivalent.
Getting some foundation in psychology of perception and gestalt theory is a good idea if you plan on doing very detail-oriented graphic work as many issues may be interacting on the same design. For example, if you fill a square with horizontal lines, it will look wider than if you fill it with vertical lines.
To further complicate things, people perceive things slightly differently, possibly based on cultural, environmental and ecological factors.
Last and most important, try to forget what you know when looking at a design and focus on what you're seeing.
Some relevant reading on optical illusions found in design here: https://blog.prototypr.io/11-optical-illusions-found-in-visual-design-295e7ae211b9
It may be one of the many reasons why a construction grid is necessary: justify the construction and position of each logo component.
If the logo is already done, try to locate construction matches. In a quick sketch I found these that surely can be improved. As an example, I didn't do any analysis of the curves radius, which can give new coincidences.
;)-- (i'm kidding!)