I was about to type “why would the engineers who upgraded search have a say on the login behavior” then I answered it myself by realizing a far more powerful search feature means they spend more compute per query. Thus it actually makes sense they use login requirement to limit the search to users who are adding actual value to their bottom line.
Again you could disagree but it is a decision that can be rooted in sound logic.
The same sound logic could have been used to put the old search being a login. That also consumed some amount of compute per query after all. Actually there's no reason to serve logged out users with anything except a login page. Why spend any resources on users who aren't adding value?
Or, maybe this is all value-destroying nonsense that only makes the service worse.
In any case, the performance of the new search system is supposed to be better, not worse. [1]
Making something more efficient but also more useful can still make it more expensive because significantly more people will now use it.
In principle they could make it all login only, but hosting open source repositories is one of the main charters of GitHub or at least it was. So it makes sense and is consistent with that vision to keep code browsing free without login. Not to mention the requirement to be available on search engines.
Code search however doesn’t seem like it’s some mandatory part of that mission, I mean all they’re asking is that you login. People in this thread seem to have a chip on their shoulder believing corporations should just give them compute for free. No they don’t!
It's amazing how you can make something explictly worse, specifically to fatten your bottom line, and have people cheer you for it.
If the goal is openness, i.e. why GitHub has the prestige it does, allying itself with open software development goals, then you should be able to do most things without having to sign up for the service. That includes browsing and searching. Neither of these need your identity.
On the other hand, if GitHub really just wants a walled garden with paying customers (either directly or the money to be made from datamining their identity and activity), it ought to shut itself off completely and get no benefit from being associated with openness.
What they've done is insidious. They're open, they're the trusted custodians of open source, but ah ah not really. For a preem experience you gotta pay up, choob! They let you search for 15 years but now they don't. And here you are cheering them rather than questioning them.
What's next? The certificate transparency project requires sign-up and logins, it's just too expensive to let anonymous users see the transparency logs you see, and by some amazing co-incidence, everyone signed in @google.com sees no results for mis-issued GMail certificates?
> Why spend any resources on users who aren't adding value?
What makes you think that logged out users aren't adding any value?
The behaviour of github here is consistent with them assuming that logged out users provide a small but non-zero amount of value for them.
The link you provided seems to suggest that the code search is supposed to be faster for users, but I can't see anything in it that suggests that it's less taxing on the backend?
> What makes you think that logged out users aren't adding any value?
I'm curious because I'm not sure of the answer myself: what value do you think anonymous users provide to them?
I suspect gating it behind a login is a simple way to rate limit, and prevent abuse with a cheap and effective way to mitigate false positives. As someone who's had to make similar decisions in the past (albeit smaller scale), I can sympathise with them turning off a computationally expensive feature that's not at the core of their product, but just an add-on.
Right, I kinda see that point, but I'm also not sure how I feel about the code search issue. On one hand, I agree that it makes the user experience nicer and can aid conversion, but on the other it's easy to abuse the feature (in a few ways off the top my head).
Some of the things you highlighted (ie the social network effects) are best experienced when logged in, and other than "just" browsing, there's not much in interaction for logged out users already (can't host code, open issues, comment, star, etc). I think allowing people to browse code and its history, alongside other parts of the platform for logged out users already feels pretty open to me
I may not have made it clear enough, but my first line was meant to be absurd and obviously wrong.
I think a service should focus on how it provides value to users, not the other way around. People use it because of what it provides to them, not because they desire to support the company's bottom line.
GitHub built a remarkably valuable product, which attracted 100M users and $1B in revenue. It has tons of public features that never required login. They could have put search, or any number of other things, behind a login at any point in the last 15 year. But they didn't, because that's not actually a good thing to do. Doing that can only ever make the feature less usable.
What drives them to do it now is unclear. Perhaps just wanting to juice their login numbers or whatever. But, performing little bait-and-switches on random features isn't going to make people use them more or spend more money. It's going to make them seek hedges against the destruction of the features they enjoy.
As for my link, the search result speed for users is directly related to cost per query, unless you believe that all of this is fake and the new search system is secretly just a bunch of expensive new hardware.
Again you could disagree but it is a decision that can be rooted in sound logic.