If your goal is naturalism, there are a lot of benefits to taking the historical approach. So much of the interesting parts of language come through historical development, especially the fun features we find more natural sounding. Some things that historical conlanging can help you with are:

Interesting phonotactics and phoneme distribution: If a language has some oddly specific phonotactic constraint, it's probably changed something. For instance, Spanish very specifically bans initial /s/ + consonant clusters and adds /e/ before them in foreign loans. Look into it and they actually added that initial /e/ before all Latin /sp st sk/ clusters, and that constraint sticks today. My Ndăkaga doesn't have any initial clusters with /j/ because any cluster of consonant + /j/ either became a palatal consonant or was simply lost.

Irregularity: What I call "regular irregularity", where some inflectional paradigms differ from the default in some particular way, come from history. Spanish "shoe verbs", where you can have verb stems alternating e <> ie or o <> ue come from vowel breaking rules you can observe elsewhere interacting with stress shifts in verb conjugations. Even less predictable irregularities, like English strong verbs, have historical sources (in this case, Proto-Indo-European ablaut).

Dialects/sister languages: If you want any dialect variation or a language family, even if you're not planning it now, you'll probably want to work historically. Mark Rosenfelder observed a couple decades ago that working forward from a protolanguage is much easier than working backward. For a small example of why you should have a protolang first: in my interview with Paul Frommer, he mentioned that, when adding a new Na'vi dialect, he had to manually go through and decide which words had /i/ vs /ɪ/ in order to create the (presumably original) distinction. If he had a protolang where that distinction had been lost, he'd just have to edit a sound change file, and he'd probably have more time to do other things with it.

So yeah, the historical method is super useful. You can do all of these things just working synchronically, but it's a lot harder and more prone to inconsistency. The caveats are that you'll need to learn how to do historical work on top of synchronic typology, because you have to also create (at least some of) the protolanguage just on typological principles. You decide what time 0 is, but you can't just not have a time 0, and we don't know enough about what the very first languages were like to start at the beginning of time (at least for oral languages).

It’s not as refined or as compartmentalized as Niklas Luhmann’s process, but art historian Victor Margolin broadly outlines his note taking and writing process in reasonable detail in this excellent three minute video. (This may be one of the shortest and best produced encapsulations of these reading/note taking/writing methods I’ve ever seen.)

Though he indicates it was a “process [he] developed”, it is broadly similar to that of the influential “historical method” laid out by Ernst Bernheim and later Seignobos/Langlois in the late 1800s.

Victor Margolin’s note taking and writing process

Collecting materials and bibliographies in files based on categories (for chapters)

Reads material, excerpts/note making on 5 x 7″ note cards

Generally with a title (based on visual in video)

excerpts have page number references (much like literature notes, the refinement linking and outlining happens separately later in his mapping and writing processes)

filed in a box with tabbed index cards by chapter number with name

video indicates that he does write on both sides of cards breaking the usual rule to write only on one side

Uses large pad of newsprint (roughly 18″ x 24″ based on visualization) to map out each chapter in visual form using his cards in a non-linear way. Out of the diagrams and clusters he creates a linear narrative form.

 Tapes diagrams to wall

 Writes in text editor on computer as he references the index cards and the visual map.

I’ve developed a way of working to make this huge project of a world history of design manageable. —Victor Margolin

Notice here that Victor Margolin doesn’t indicate that it was a process that he was taught, but rather “I’ve developed”. Of course he was likely taught or influenced on the method, particularly as a historian, and that what he really means to communicate is that this is how he’s evolved that process.

I begin with a large amount of information. —Victor Margolin

Dry farming, a method that’s been used for centuries to grow grapes, almonds, and olives in Mediterranean countries, requires soils with enough structure to hold moisture from seasonal rains for months at a time—in California, these rains happen between October and April. One method is to plant young vines that are grafted to vigorous rootstocks relatively far apart and water them for only their first two years in the ground. The point is to encourage their roots to dig deep into the dirt from which they’ll pull stored rainwater starting in year three.

“In dry farming, you’re putting resistance into the system,” says Stephen Gliessman, an emeritus agroecologist at the University of Santa Cruz who also co-owns the dry-farmed vineyard Condor’s Hope in the Cuyama Valley of northern Santa Barbara County.

Though plenty of wine grape growers in the state practice dry farming, the method represents a drop in the bucket of a $70 billion business. Tightly spaced, high-yield, drip-irrigated vineyards are much in favor; their practices encourage roots to hang out near the surface of the soil, where they expect to find water—and they can’t survive without a frequent fix.

I think we could all agree that if you had never been exposed to mathematics in your life, you'd be setting yourself up for failure if you began your education by jumping straight into differential equations and Fourier analysis without first developing a broad foundation in basic arithmetic, algebra, geometry, calculus etc.. We generally accept that there are prerequisite skills that we need to develop before trying to tackle more complex and advanced problems. Could you do it?  With enough motivation, raw intelligence, and a lot of time spent back-tracking, you might be able to work your way through a problem here and there and even arrive at a correct answer every so often, but your overall understanding of the material, your ability to perform consistently and to apply those skills broadly would be severely compromised. Yet we as a community of enthusiasts and hobbiests don't think twice about trying to start with the more 'advanced problems' as we embark on our journey learning about arms, armor, and other aspects of historical material culture.

1. Minding the Gap between Event and Gospel