You said you try your best and still fail. Here is why: you have been copying the surface. This is the machinery underneath, measured rather than admired - every sentence he writes, counted, and turned into something you can practise for eight hours a day.
Each of these contradicts the standard advice. Each is measured from his actual prose, not inferred from reading it.
Memoir to memoir, 2014 to 2024, mean sentence length rose from 13.6 to 14.1 words. Sentences over 30 words nearly doubled.
And the famous "long sentence, then a short one" rule is false. Measured across 91,862 sentences: after a 30-word-plus sentence the next averages 17.7 words against an overall mean of 14.6. It gets longer, in all 15 books, no exceptions. Long sentences cluster. What is real is the spread - he writes a much wider range than you do, not a tidy alternation.
Mean word length across a decade: 4.25 characters, then 4.25 characters. Identical. Words of 9+ characters held steady at 6.4%.
What collapsed was Latinate abstraction - the -tion/-ment/-ness/-ity layer, down about 35%. He did not simplify. He substituted: named objects, clock times, proper nouns.
Across 528 posts, 2011 → 2026: "you" fell from 48.3 to 28.0 per thousand words while "we" almost quadrupled. Early Om Swami preaches. Late Om Swami testifies.
But this is blog-only. His books went the other way - "we" peaked at 9.98 per 1k in 2019 and fell to 1.91 by 2024. In books, genre sets the pronoun, not the year: his 2014 memoir runs 43.9 "I" per 1k, his 2024 memoir 44.4. Do not copy the "we" into one-shot writing for strangers; it is priced against a fifteen-year relationship.
Newest on top, oldest at the bottom, as you asked. Fifteen of these are on your Mac and were measured in full. Two are missing - buy those, because the study is only as good as the corpus.
The Book of Kindness (2019) and A Fistful of Wisdom (2017). Buy direct from os.me/books so the money reaches the ashram rather than a reseller. Drop the PDFs beside the others and the whole analysis can be re-run against them.
Twelve books plotted oldest to newest. Watch the second chart especially - that is the one that overturns the "write shorter sentences" advice.
Any claim here that could not survive a re-measurement has been changed rather than quietly dropped. Three did not survive.
The first version of this study said the long-then-short beat was his governing rhythm. It is not. Measured across 91,862 sentences, the sentence following a 30-word-plus sentence averages 17.71 words against an overall mean of 14.61. It gets longer, and it does so in all fifteen books with no exceptions. Long sentences cluster. The true verdict beat, a 25-word sentence answered by one of six words or fewer, occurs in only 1.6% of his sentences: a deliberate device, not a metronome. What is genuinely distinctive is the width of his range.
Caught by your own earlier study of his Guru Purnima address, which had already found and corrected the same error on a different corpus. Two independent corpora, same conclusion. om-swami-prose-study.vercel.app
Initially written up as a universal evolution. It is not. His books moved the opposite way, with "we" peaking at 9.98 per 1k in 2019 and falling to 1.91 by 2024. In books, genre sets the pronoun rather than the year.
Averaging his last forty posts masked both the em dash going to zero in the last eleven and his 2026 sentences tightening to about 14 words from 16.6. Recency had to be measured separately, not averaged in. The Workbench bands are now calibrated to the last twelve posts rather than to the books.
In the order that works. Evolution first, because it tells you which habits to drop before you try to add anything.