So, fake smallcaps are bad, and SE doesn't have real ones. Do we want real ones? I assume the increased page lag would be comparable to the overhead increase we got when we implemented MathJax (or less than that). Also it would take the dev team time to implement, which they could spend on other stuff. Do you guys think it's worth asking for it to be implemented? (Related question on Meta Stack Exchange.)
Small caps won't be implemented at SE any time soon because they can't be represented in the flavour of Markdown used by SE currently, and all feature requests that require extending the Markdown parser, which aren't mission-critical, have been turned down. (The one mission-critical extension SE has done is spoiler blocks, and they're very buggy.)
The current Markdown flavour that SE uses is slated to (eventually) be replaced by something that implements the CommonMark flavour, which is a better behaved specification of Markdown. However, small caps doesn't appear to be one of the features included in CommonMark currently.
This lack of support is almost certainly due to small caps being styling rather than semantic content, like font colour is a matter of style rather than meaning. Small caps also have no representation in HTML for this reason: to get small caps in an HTML document requires a designer to decide that certain semantic parts of a document should be conveyed by small caps, using CSS styles.
Since Markdown is designed to represent only content and is agnostic about style, very likely small caps will never be supported by any flavour of Markdown SE uses.
(This all despite the fact that, personally, I would love to be able to use small caps here. I just know that matters of styling aren't going to find their way into Markdown.)