Two hundred posts that barely link to each other, does internal linking actually do anything, and how without it looking spammy?

It is real, and it does two distinct things that are worth separating because they suggest different tactics.

Discovery and crawling. A crawler finds pages by following links. A page nothing links to is reachable only from a sitemap or an external link, and is crawled rarely. If you have pages that never seem to get picked up, this is frequently why.

Meaning and importance. Links tell a search engine which pages are related, what they are about (through the link text), and which of your pages you consider important. A site where everything is an island gives none of that. You have four years of topical authority spread across two hundred disconnected pages, and nothing telling anyone they belong together.

So yes, retrofitting is worth it: probably one of the highest-return things available to a site in your position, because the content already exists and this is purely arranging it.

On the spam worry: the sites you are thinking of link because they are linking, with generic anchor text, many times per paragraph, to whatever needs traffic. That is not what this is. Three to five links per article, placed where a reader would genuinely want them, is invisible as a technique and does the entire job.

Where to start at two hundred pages, because doing it in publication order will take months and get abandoned.

Find your orphans first. Crawl your own site with any of the standard tools and list the pages that no other page links to. Those get the largest benefit from a single link and they are usually a surprisingly long list.

Then work top down by value. Take your ten or fifteen pages with the most traffic and add links from them to related pages. Those are the pages with the most to pass on, and it is the fastest way to get the rest of the site connected to something strong.

Then group by topic. Find the clusters that exist naturally in two hundred articles, you probably have five or six subjects. Link within a cluster generously and between clusters sparingly. That structure is what tells a search engine you have depth on a subject rather than two hundred unrelated pages.

Consider one hub page per cluster that links to everything in it, with the articles linking back. Cheap to make and it gives each group a centre.

Realistically that is a couple of days of work for two hundred pages, not weeks, and the first two steps are most of the benefit.

On doing it well rather than merely doing it, four things that matter:

Anchor text should describe the destination. Not click here, not this post: the words a person would use for what is on the other end. This is the part that carries meaning, and generic anchors waste the link entirely.

Vary it naturally. The same exact phrase every time looks manufactured. Write the sentence and let the link fall where it fits.

Place them where a reader would want them, which is usually mid-article at the moment a related idea comes up, not in a block at the bottom. Related posts modules at the end are fine and they do much less than an in-text link.

Link both ways when it makes sense. New article to old is the habit everybody has; going back and adding old to new is the half nobody does, and it is where the value is for a four year archive.

One more thing worth doing while you are in there: fix broken internal links and redirect chains as you go. Four years of a site almost always has some, and they waste the exact thing you are trying to create.

And build the habit going forward, every new article links to two or three existing ones and you add one link to it from somewhere older. Five minutes per post and you never have this problem again.