I think it is a good time to remind everyone of the proper context for using different tools to avoid algorithmically boosting bad content. This is important for all skeptics, because the very act of linking to something you are debunking can make it more visible on platforms like Facebook and Google.
One of several tools for this purpose is DoNotLink. There was a minor kerfuffle last week in which the Food Babe website unsuccessfully attempted to block incoming links using DoNotLink. That raised the potential that skeptic reliance on that service might have disadvantages.
I’ve also noticed that in addition to many people on social media who’ve adopted DoNotLink, some bloggers are also using it for links within their posts. Frankly, this is overkill and I don’t recommend it. There’s already a standard HTML feature for handling this on web pages – it is called NOFOLLOW. In this post I’ll compare the two and offer advice on when to use each.
HTML Links Can Be Disavowed With NOFOLLOW
I’ve written about the rel=NOFOLLOW attribute on two different occasions. It is an attribute you can add to a hyperlink in any context that allows HTML editing – which means most any blog or web editor. In the current HTML5 specification it is described this way:
The nofollow keyword indicates that the link is not endorsed by the original author or publisher of the page, or that the link to the referenced document was included primarily because of a commercial relationship between people affiliated with the two pages.
(Emphasis mine). The idea of not endorsing a link is exactly what we are after here, thus NOFOLLOW is the perfect tool. As part of the HTML standard adhered to by all web pages, it is uniformly supported across different tools and services. It is unlikely to break. Let’s compare that with Do Not Link.
Disadvantages of DoNotLink
I’ve been a big booster of Do Not Link, but I must be honest – there are disadvantages to it. One obvious one is the usual way of using it (like a URL shortener) obscures the link destination. In these days of scams, phishes and malware many users are justifiably wary of clicking mysterious looking links.
You can work around this by using this format for your links to make their destination obvious, as it is still supported:
But even if we use this link format, there’s another disadvantage: what if Do Not Link goes away? Unlike major URL shortening services like Bitly, this service is not supported by a well-funded company. Its creator Menno Van Ens could decide to move on and drop support for the service, in which case all our links would break.
In the social media context, this is actually not a problem; social media posts are intended to be short-lived. But broken links on the web are a huge problem already, we don’t need to add to that. So this should guide our choices here.
Can We Dodge Referrer Blocks?
A further question is raised by the Food Babe’s actions last week – what if other pseudoscientific or misinformation websites attempt to selectively block incoming skeptic links in a similar fashion? Are there proactive measures to prevent this? As it turns out, there are.
Food Babe was using something called a referrer to detect and block DoNotLink. As it happens, HTML5 has added a new attribute to selectively drop the referrer on a link-by-link basis. It is called (logically enough) NOREFERRER. (Note: the word “referrer” is spelled correctly here, unlike in the equivalent HTTP header).
The new attribute looks and works just like NOFOLLOW. Here’s what it looks like on a link, by itself or combined with nofollow:
<a href="http://google.com" rel="noreferrer">link one</a> <a href="http://google.com" rel="nofollow noreferrer">link two</a>
HTML5 is a new standard, so this is not universally supported yet. But as of this writing Chrome, Safari, MobileSafari, Opera and Firefox all support it. You must be running an updated version in most cases – which is a good idea in any case. Microsoft’s Internet Explorer does not yet support this. (If you technically savvy and are concerned about lack of universal support, check out noreferrer.js on Github, which claims to solve this).
I’ve not yet seen any evidence of other use of referrer-based blocking against skeptic links. But another advantage to blocking the referrer would be to remove some incentive to create clickbait specifically targeted at skeptic bloggers. Removing the referrer data would make it a bit more difficult for the perpetrators to use analytics to see their tactics had actually worked as intended.
And so, taking all of that into account, here are my current recommendations for best practices for linking without boosting the algorithmic scoring of misinformation:
- On web content (websites or blogs) which you create, add rel=”NOFOLLOW” or rel=”NOFOLLOW NOREFERRER” on all links to websites containing misinformation.
- In comment sections or forums, check whether the site automatically add NOFOLLOW to links before posting – most do. You can check using View Source or Show Source in your browser. If you don’t see NOFOLLOW on links in comments, don’t post one – you’ll probably not be able to add the REL attribute yourself.
- On social media (Twitter, Facebook, et.al.) continue to use DoNotLink or the RBUTR toolbar for links to misinformation. Both support the format documented above for self-documenting URLs, if desired. They prevent social media algorithms from treating your posts as endorsements, just as NOFOLLOW does for HTML.
Please discuss in the comments.
Wonderful and informative (as always). Thank you Tim!
Pingback: Critiquer un site sans le soutenir | Bunker D