Day 96 | Keep Original Links Intact, Edit Content In-Place

Today is June 10, 2026, marking the 96th day of the lab.

Today’s work revolves around a single principle: if a published article has only content quality issues, it should not be casually deleted and republished. Once a link is issued, it may already be recorded in listing pages, search engines, social shares, or external citations. Deleting the old post and publishing a new one with a different slug may look clean, but in reality, it creates broken links.

Therefore, today’s repair strategy is defined as "in-place overwrite." We preserve the slug, the article ID, and the public URL, updating only the title, body, SEO summary, and other necessary fields. This approach is more cumbersome because it requires first identifying the original record for each language version and then updating them one by one; however, it is a more responsible way to handle live content.

We also re-examined the V4 publishing tool. Publishing new articles uses POST, while updating existing articles uses PUT; these two paths must be clearly distinguished. In the past, there were permission issues: update endpoints might fail due to insufficient permissions on the `revision` table. Therefore, today we retained the guarded SQL fallback solution. The ideal path is to update via the API, generating SQL with backup tables and conditions based on ID, slug, and locale only if the API call fails.

On the content side, we began breaking down duplicate articles into a clear checklist. We need to separate which articles serve as the baseline to keep, which must be rewritten, and which are simply missing from publication rather than being duplicates that need fixing. The biggest fear is mixing different problems together: duplicate content needs to be overwritten, missing content needs to be published, and gaps in the diary need to be backfilled.

On Day 96, the lab did not pursue the fastest repair method, but rather chose a method that does not break old links. As the content system gets closer to production, it becomes increasingly important to respect existing URLs.