This makes me wonder if they're retiring sites.google.com any time soon?

But otherwise I think it's about traffic and tracking.

First in that you can track visitors to the short link, perhaps in a more centralized place if your org has multiple systems or if you're linking to sites you don't otherwise control. And secondly it's used when tacking on tons of query parameters that would otherwise make it too long even with expanded character limits. And just aesthetically a short link might be more aesthetic in your post.

Imagine seeing "link.short/abcd1234" vs "linksrus.corporate/blog/2025/08/12/the-title-of-our-post.html?utm=something&some-other-tracking-param=abcdefghijklmnopqrstuvwxyz&even-more-data-in-the-url&some-hashed-value=QXNoIG5hemcgZHVyYmF0dWzDu2ssCiAgIGFzaCBuYXpnIGdpbWJhdHVsLApBc2ggbmF6ZyB0aHJha2F0dWzDu2sKICAgYWdoIGJ1cnp1bS1pc2hpIGtyaW1wYXR1bAoKT25lIHJpbmcgdG8gcnVsZSB0aGVtIGFsbCwKICAgb25lIHJpbmcgdG8gZmluZCB0aGVtLApPbmUgcmluZyB0byBicmluZyB0aGVtIGFsbAogICBhbmQgaW4gdGhlIGRhcmtuZXNzIGJpbmQgdGhlbS4KCg"

Of course, this is possible in other ways, but note that I said, “easy”. ;)

Instead, they're just disappearing _all_* goo.gl short links. The overwhelming majority of which are benign links made by users who were promised a super stable URL link shortening service backed by the Google brand.

*edit: Not all, but nearly.

For goo.gl links that were created by google, continue redirecting them as normal. For others, show a warning page explaining to the user that the link wasn't created (or vouched for) by google. If they press an "agree" button, still don't show a clickable link, but instead show it as plain text to be copied.

The goo.gl link shortener hasn’t accepted new links for many years. Over 99% of the links had no recent activity. The play was to scrape the web for old goo.gl links that went to expired domains, register the domain, and then you have a goo.gl URL that you can send wherever you want, indefinitely.

Nearly all of the angry blog posts, Tweets, and HN comments missed this and jumped to the conclusion that it was purely a cost cutting measure, but link official-looking open redirect URLs are a big deal in the security space.

It's mostly for print usage anyway.. and with all the TLDs out there now, you can definitely still get some short options. I literally have one strictly for (reverse)dns for my server(s). Another for my mail server, separate from the domains the server hosts.

Also, hosting your own means you don't have to worry about being associated with externalized spam links. Of course, if you are a spammer, then it works even better for users that will block you.,

Later, adding things like analytics and tracking (eg: not just in social media, but also in email campaigns) became another reason to use them, especially for those less tech inclined.

They could possibly provide a GCP service where you make an authenticated request to look up the value of a given goo.gl key. That would mitigate fishing concerns, eliminate the pressure of running a productionized legacy service, and allow the to do use quotas etc to tamp down on abuse. But that also would be covered by the regulatory laws and I don't know what they say about such a thing.

Two weeks ago [1] they were doing 37k a minute with ETA just barely before end of the month. Now it's ~55k a minute, and ETA of just 5 more days.

[1] https://news.ycombinator.com/item?id=44688478

Also, don't use link shorteners... and don't build link shorteners as it just enables people to use them.

They already did this in 2019.

The service has been deprecated for a very long time.

> Instead, they're just disappearing _all_ goo.gl short links

This is false. They are sunsetting inactive links.

> All other [active] goo.gl links will be preserved and will continue to function as normal. To check if your link will be retained, visit the link today. If your link redirects you without a message, it will continue to work.

https://developers.googleblog.com/en/google-url-shortener-li...

"Recent" is defined within the last year. If the Wayback Machine adopted this logic, it would be useless.

The security concerns were largely addressed by not accepting new links. This was a cost cutting measure, plain and simple. I think we all agree that a goo.gl shortener was a terrible idea to begin with, and my blog post even shows evidence that folks knew this was a bad idea at launch.

Giving the domain to a 3rd party is not going to happen.

It would be a serious breach of trust for them to publish the database. It likely includes links to non-public YT video URLs, for example.

But they actually backtracked and said they'll keep the "active" links working.

Why even spend the effort to remove the "inactive" links? They must feel they represent some sort of liability?

Or would it have been too embarrassing to just cancel the whole turn-down plan?

it is relatively simple and inexpensive to operate and there is no reason it can’t continue operating into the indefinite future.

I disagree with the op on one major point… They suggest that the proper way for a service like this to operate is with a public database of links…

However, when I think of all of the different ways, one might use a shorter, it seems obvious that many links would necessarily be sensitive and/or private.

Therefore, I consider myself to have a dual mandate: privacy of the links created, and a duty, that I have assumed, of running the service for decades.

It is true that it is not _all_ links, apologies. "Inactive" here is defined as "not visited in 2024" which is a crazy small envelope. I wouldn't be surprised if nearly all links were deleted.

...which is even harder to justify than removing them all.

It would make sense if they were pruning links whose TARGETS were no longer responding. But all the unused links are costing essentially nothing. Essentially all the cost was spent already.

> The play was to scrape the web for old goo.gl links that went to expired domains, register the domain, and then you have a goo.gl URL that you can send wherever you want, indefinitely.

It's absolutely not rare. I can't tell you how many times I sat in the back of a classroom, squinting at a whiteboard, trying to type copy a URL that looked like this: https://www.nytimes.com/2025/08/12/technology/silicon-chips-...

I do think that, however, that a service can keep their DB public _iff_ users know that's the case before creating links. Often times, folks are just shortening public links, but it is true that people sometimes share "private" links (e.g., Google Doc private links, etc...) which shouldn't be enumerated.

[0] https://archive.org/details/301works-faq

Those links smack of the same mindset behind "Refactor to rewrite procedure A in terms of procedure B. 7 lines deleted; 17 lines added."

This NYT link is actually pretty good. year/month/day/category/title is pretty nice. The query parameter ruins it, but that's kind of unavoidable if you want a reproducible link with extra data. Usually you can omit query parameters and the URL should still work.

0: https://wiki.archiveteam.org/index.php/ArchiveTeam_Warrior#A...

1: https://github.com/ArchiveTeam/warrior-dockerfile/blob/maste...

Is it because they're worried that the domain name goo.gl in the link implies a Google endorsement? Seems like they should have thought of that before launching the service in the first place?

Still, the frequency of actual abuse must be low and going down over time (due to the data set being read-only since 2019 and actual traffic to these links surely decreasing as time goes on)...