> While most features of goo.gl will eventually sunset, all existing links will continue to redirect to the intended destination.
And
> After March 30, 2019, all links will continue to redirect to the intended destination.
[Emphasis in original, in bold!]
It is hard to read this as anything but saying the continued serving of redirects will _not_ eventually sunset, right? While most features will eventually sunset, all existing links are not most features, and will continue to redirect.
Clearly they changed their mind. I mean, it's not shocking, especially from Google.
It would be decent and transparent of them to admit it though. Yeah, we said that we were going to continue to redirect existing links, we changed our mind, sorry about that.
This is Google we’re talking about. I think at this point they are institutionally committed to the notion that their users ought not trust them to keep services going and if they do, then shame on the user.
How much would it cost Google to just keep a few hundred webserver replicas running on some Borg cluster somewhere that do nothing but match and redirect incoming requests to their destinations for a few more years?
Am I underestimating the complexity of this at Google's scale?
It's because of the constant churn in Google's shared infrastructure.
I've been the maintainer for production services left orphaned by departing ex-Googlers. They usually go down within 3-6 months of their previous maintainer departing - either there's some PCR because the cluster they're homed in is going down for maintenance or transferring to another datacenter, or their dependencies migrate to a new RPC format, or somebody sticks a new quota on internal requests for a service, or your PA loses machine budgets, or your RPC deadlines are now more than the average latency for that service. It's stupid stuff like that - assumptions that worked a few years ago don't work anymore when the infrastructure is a moving target.
Google's code is immensely complex, so once the institutional knowledge goes out the door, debugging and fixing problems becomes very difficult. Your maintainer, once the original author has left, is probably doing this as an unfunded 20% project; she will get zero credit on her performance review for making sure that a service her VP doesn't care about stays up. So it's usually easier to do nothing and just let it start failing. Announcing a turndown is actually a courtesy in this case.
I am at a different large company. We got a pile of alerts for a service's staging environment being unused. Of course it's unused, we haven't done a release in months.
We clicked the appeal process on those alerts. You could tell from their response that this was a novel concept.
> Google's code is immensely complex, so once the institutional knowledge goes out the door, debugging and fixing problems becomes very difficult.
How complex can a static (since there's no new URLs being added) URL shortener get? If it's really that bad just rewrite it to something simpler in an afternoon and be done with it. It must be a better option than the bad will of sunsetting something like this will generate no?
More than $0. There is no consequence to turn down (and sunk cost of labor doing the work), cost to keep running is above $0. If the argument is good will, the value of good will to current state Google appears to be $0.
Throw a few dollars at the Internet Archive [1], ask ArchiveTeam to add the Google link shortener to the Warrior archiver [2] [3], understand why to avoid Google in the future.
The cost to keep a single server which only responses with 301 redirection is worth pennies, or 5 minutes of Googlers' earnings. Breaking the global web is much more costly.
Few hundred webservers? It could be a single VPS with a single script just looking up the alias in database (possibly SQLite) and serving the 301 redirect.
Print publishers might have a case for URL shorteners but they should own and maintain their own instead of a 3rd party service they can’t control or prevent from disappearing.
Plenty of articles in the Communications of the ACM magazine use 3rd party shortened URLs and it’s unacceptable. They should know better and should host their own link shortener via their digital library system.
I remember when Tweeple were inventing hashtags and retweets in Twitter textboxes, bit.ly was the saviour that prevented url deadweight taking up our precious 140 characters. If 2010 Twitter allowed [markdown urls](https://example.com/longlonglonglonglonglonglonglongurl) and counted only the visible words, maybe the entire link shortener industry would have died off.
From a security perspective, this is a good idea. People intuitively trust Google links. URL shorteners hide what you're clicking on. Sometimes even informed people click on links.
I remember a war between Users and IT - the shitty CMS everybody used had a crazy-small limit for links, so people wanted to use URL shorteners, which then got blocked for security reasons ...
I have seen linked.in shared as if its job posts shared on LinkedIn, then they redirect to some other url. A job seeker usually has to visit another url to apply to a job posted in LinkedIn, so its not completely suspicious.
just out of curiosity I wanted to check whether I had some goo.gl links that I might want to migrate but apparently goo.gl now redirects to this announcement.
Would be happy to know if there is a way to access goo.gl links created while logged in with my account.
Normally a product shutdown has a contained set of affected users, but this (incredibly computationally simple) service getting turned off will linkrot a noticeable fraction of the internet. This is shittier than usual.
Late to the discussion but as a former Googler I'm embarrassed for the company they didn't figure out a way to keep these working forever. They really just don't give a damn over there any more, do they?
My dumb architecture suggestion for long term resilience: replace it all with absolutely static HTML files with a redirect in them in meta tags and Javascript (and HTML). Ugly but can be served forever on anything that can serve a web page.
"We understand the transition away from using goo.gl short links may cause some inconvenience." ... Hum, ya ... Non tech people might have made lots of links in legacy internal doc's, email conversations and other reference maternal ... Maybe even a thing called "Paper" ;-)
Another one for the Google Graveyard I guess. Feels like everything outside of a few core products is either being shut down or abandoned these days...
One could probably write a browser extension that catches clicks on goo.gl/ URLs and queries a web backend for the original URL (a backend that needs to be filled with the original short-to-full URL mapping , and so preserve the ugliness.
And since we're all obsessed about money, one could put a 5 second ad before doing the actual redirect, and offer a pro subscription for redirects for many bucks a month...
A little earlier: the DoubleClick merger happened in 2007 and it took a little time for the ad guys to end up running everything, coincidentally also marking the end of their ability to launch a successful product.
I’m sure it was related - everyone got really focused on the bottom line then – but I think they also had the problem that ads were so successful that it became hard to even consider other models because ads just instantly generate money, and a lot of their attention seemed to focus on that sit-and-consume model rather than, say, getting serious about building tools people would subscribe to.
G-Suite is kind of sad that way because there was some cool technology and initial goodwill but they just kind of coasted and slowly gave the market away to Microsoft, I think because the fountain of ad dollars meant nobody was worried about their bonus being cancelled or their shares going down significantly.
> While most features of goo.gl will eventually sunset, all existing links will continue to redirect to the intended destination.
And
> After March 30, 2019, all links will continue to redirect to the intended destination.
[Emphasis in original, in bold!]
It is hard to read this as anything but saying the continued serving of redirects will _not_ eventually sunset, right? While most features will eventually sunset, all existing links are not most features, and will continue to redirect.
Clearly they changed their mind. I mean, it's not shocking, especially from Google.
It would be decent and transparent of them to admit it though. Yeah, we said that we were going to continue to redirect existing links, we changed our mind, sorry about that.