Would HTML/Unicode formatting cause troubles?

Question

Although, as of now, this comment collected 17 upvotes, I still hope that the thought was expressed not precisely enough. (Mis)typesetting like

     b^2 - a^2 = (b+a)x(b-a)

can induce vomit, many would agree. But is

     b² − a² = (b + a) × (b − a)

very different from $$b^2 - a^2 = (b+a)\times(b-a)\ ?$$

In specific cases, there may be some arguments against using LaTeX, ranging from avoiding unnecessary complications ($1,729$ renders with an annoying space – do you know which code produces $1{,}729$?) to aesthetisation (“a 2 × 2 matrix” does not protuberate unlike “a $2×2$ matrix”, and I prefer 2π over $2π$, as well as 360° over either $360°$ or $360^\circ$; the last one is also not searchable due to incorrect code point). Also, it is a common practice to avoid LaTeX in titles; although they haven’t HTML, Unicode is available.

Is here some rule or prejudice against HTML/Unicode-formatted formulae? Ones similar to those I produce at various Internet sites, namely: StackExchange, Wikipedia and, very recently, Quora. Have LaTeX codes some importance but as one of formatting mechanisms?

---

Updates:

 1. I found that ISO 80000 advises against italicizing symbols of constants. Namely, ISO deems that π (for the area of unit circle) is incorrect whereas π is correct. I doubt there is a convenient way to display roman-type Greek lowercase letters, namely α β γ δ ϵ ε ζ η θ ι κ λ μ ν ξ ο π ρ ς σ τ υ ϕ φ χ ψ ω, with either T_EX or MathJax.
 2. A newer discussion about questionable MathJax practices, borrowed from early days plain T_EX kludges, is currently hosted here.

Answer 1

Although there are some great comments above, one thing no one has mentioned is people who use assistive technology like screen readers. Having the mathematics clearly marked as mathematics (as oppose to "fake" mathematics using other HTML tricks) makes it much easier to properly voice the mathematics for screen readers. Although few screen readers currently handle mathematics, the number are growing, and the MathJax team is working with several to help improve the situation. The format used by most screen readers is MathML, and since MathJax can generate MathML from your TeX input, that means that any math that is entered in MathJax notation will be accessible to those users with special needs. The upshot is, if you use MathJax to mark up your mathematics, even when it is possible to do it using HTML, it will help out those readers who use assistive technology.

A related issue is that people who wish to reuse the content of the questions and answers, not just in LaTeX papers, as has been mentioned in the comments, but for other machine processing of the math (math aggregators, math search, analysis of mathematical content of questions, copying and pasting into other programs, etc.), will find it much easier to identify the mathematical sections of the questions if they are clearly marked as mathematics rather than marked up in HTML in other ways. MathJax can create a good machine-readable form (MathML) for the math content when it is written in TeX, and that can be used as an interchange format for other programs. For example, the MathML obtainable from MathJax's "Show Math As..." contextual menu can be pasted into programs like Mathematica, and even Microsoft Word, while retaining its mathematical meaning. That simply can't be done with HTML-formatted mathematics.

My own feeling is that anything that improves the semantics of the content (without unduly burdening the author) is a good thing. Using a mechanism (like MathJax notation) that clearly delimits the mathematics from the other text on the page is an important example of this.