I’ve been watching a lot of the actions that StubBot’s edits partake to. However, I also have been seeing some edits by others saying that some of the stubs it places aren’t really stubs - it’s only getting worse with time. How does StubBot find articles to stub? How does it determine what could be a potential stub and what couldn’t be a potential stub? Does it utilize the NAB score? I’m perplexed and wondering! There are so many inaccurate stubs that it’s leaving me bewildering! I know we’ve been “panning for gold” but some of these articles that aren’t really stubs at all have been marked as stubs, and should’ve been marked as something else for further edit work.
4 Likes
I’m not sure about the bot’s algorithms, but I know it can be told to ignore pages if they meet some criteria or another - for instance, expert co-authored articles. But I also saw Chris H demoting a lot of articles in the Health category recently, which included a lot of co-authored pages, so I think the most recent Stub Bot sweep may have been part of a broader move to deindex health articles that could be harming our position in search results. I’ll ping @JayneG
to confirm the actual reason, though, because that’s just a guess on my part.
There are a bunch of articles within tech articles (a lot of the non-stubby stubbed messes) that it’s been grabbing that have as much as they could possibly need (and then some, for some of them)), and some just need other edit work, and stub templates wouldn’t really help.
Tech articles are more likely to get mistakenly hit by the bot, because they usually don’t have as much content as other articles; if they’re complete, you can just unstub those. But to my understanding, the bot is meant to check if a page needs deindexing, and doesn’t check for other needed work. In that case, I’d imagine it’s fine to replace the stub with a more appropriate tag.
1 Like
There’s no single metric that determines whether an article will be stubbed by Stub Bot. I tune the approach to address different issues depending upon our needs and best practices for SEO. Some of these are obvious, e.g., short articles with bad helpfulness ratings. Other times, it’s processing articles that may be hurting wikiHow’s quality ratings in Google. For example, when we have an article on a popular topic, but is not reaching any readers, that is a strong sign Google considers that article to lack authority. Having many such articles can hurt wikiHow’s domain reputation. This is why you may see Stub Bot tag articles that seem comprehensive.
Lately, we’ve also been looking for topic areas that are simply a poor fit for wikiHow, which is why I’ve demoted a bunch of medical articles. Although we have good reasons to know our quality is high, it’s difficult for wikiHow to be considered a leading website for serious medical topics, when compared to specialist sites that do nothing but medical topics.
The end goal of all of this work is to get helpful information in front of as many people as possible, and this process helps the best, and most relevant content shine. I liken the process to pruning trees—by removing some of the less healthy branches, you help the whole tree to grow and thrive.
10 Likes
Thank you for clarifying, @Chris_H
