Every item database faces the same temptation. You have data on 150 fruits or 200 pets, the template generates a tidy page for each one, and every page is another lottery ticket in search. The industry-standard move is to publish all of them and let Google sort it out. We do the opposite: an item page on this site starts life invisible to search engines, and it stays invisible until we've written enough about the item to justify a stranger's click.
The threshold is concrete: 250 words of real, item-specific writing - what the item does, how you get it, what it's actually good for, how its market behaves. Under that line, the page still exists for humans (it has the data card, the fair-trade comparisons, the working links) but carries a noindex tag. Over the line, the tag comes off automatically at the next build. Right now that split is 80 pages invisible, 135 indexed. The invisible set is mostly Blox Fruits commons and Grow a Garden crops whose honest writeup would be a paragraph of the obvious.
The trap we're avoiding
A thin item page isn't neutral - it's a small lie about what's behind the click. The search snippet promises "Barrier fruit: value and guide," the visitor arrives, and there are ninety words wrapped around a stats table they could have gotten anywhere. They bounce. Multiply by a hundred stubs and you've taught both your visitors and the search engine the same lesson: clicks here get wasted.
The sites that carpet search with a page per item are optimizing for the lottery, not the visitor - every stub a ticket, substance optional. It works until quality systems catch up with it, and the graveyard of deindexed gaming databases suggests they eventually do. But honestly, even if it worked forever, we wouldn't want the traffic. A visitor who bounces off a husk was never going to come back for the tools, and the tools are the point of this site.
How a page earns its way back
The mechanism matters more than the sentiment, so here it is. Item prose lives in our data files alongside the values. When an item's written fields clear the threshold, the next build removes the noindex tag, adds the page to the sitemap, and wires it into the site's internal links - no editorial ceremony, no exceptions committee. The rule runs in code, which means we can't quietly grandfather in a favorite.
We work the queue in tranches, highest-value items first, because that's where reader demand actually is: Kitsune got its shrine route and market history written before any common ever will. Some pages may never earn their way in. A starter fruit whose complete honest description is "cheap, weak, eat it for the mastery" does not need to be in Google, and pretending otherwise would just recreate the husk problem with 250 words of padding - which is why the rule is a floor for earning the writeup, not a target to pad toward. If we can't write 250 genuinely useful words about an item, that's information too.
The same rule, everywhere we could apply it
Once you see coverage-versus-substance as the axis, most of this site's odder decisions turn out to be the same decision. The limiteds tracker lists hundreds of modern collectibles as "value pending" rather than inventing numbers for them. The trade checker refuses verdicts inside its own error bars - the reasoning is in the even band. The codes pages carry dated verification stamps instead of claiming daily freshness they can't prove. In every case the alternative was more apparent coverage, and in every case the coverage would have been hollow.
There's a practical bet underneath the principle. We think a site where every indexed page survives a skeptical click eventually beats a site with three times the pages and a bounce problem - in search, in return visits, and in the only currency that compounds: whether people trust the next number they see here. It's a slow bet. Slow is fine. The fruits aren't going anywhere, and neither are we.