Hacker News new | past | comments | ask | show | jobs | submit login
GitHub backtracks on deprecating the trending tabs page (github.com/orgs)
153 points by TowerTall on Sept 3, 2022 | hide | past | favorite | 65 comments

I do find it a little odd all the comments in this thread are focusing on it as though it was a straight telemetry/metric decision.

From the update, this feels more like it was problem area for maintenance purposes, then had its telemetry assessed for decision.

"The data pipeline that powers the page has historically proven problematic to maintain and so given the low usage, we decided to switch off the Trending page."

The maintainers likely considered the resources and effort in maintaining it were too high. When they looked at the telemetry and saw it wasn't utilized considerably, they assessed it was better to cut it, than keep it or create an alternative.

If it didn't have a cost associated with it, it would likely have never become a target.

My read of that line is, “the person who built this feature left, no team owns it, it needs a full rewrite, but we have other priorities”. It’s not surprising the metrics show the page usage isn’t growing. GitHub’s current UI doesn’t really direct anyone to it. What their metrics don’t show them is how important the page is to the users who do use it, and how important those users are. But it seems they have found out.

Since Microsoft acquisition GitHub quality seems in decline, constant outages, git pushes that doesn't goes through while status page is green and now the attempt to remove trending page...

Is it the end of the original GitHub's spirit as a social coding?

Not sure what you mean by "quality is in decline".

I'm pretty happy (much like others) with GitHub now more than ever. Github has been releasing new features left and right.

Some of them are an unwelcome sight (think Github Badges) while others (think Github Issues) are a super productive enhancement imo.

Honestly, I don't really see it. GitHub's availability has never been great and it doesn't really seem worse now aside from having more features to go wrong.

I haven't seen much change at all since the acquisition personally. Probably the only thing I can pick out is their dropping of Atom in favour of vscode.

What have you specifically seen change since the MS acquisition that's likely from MS and not just GitHub?

I don’t agree with this statement. Im happy to see the innovation that have come out from GitHub in the last years.

What im really missing and hoping for is an even more open and extensible community platform. Discussions is a good start, it would love to more innovation there.

The initial call to deprecate this page was quite obviously driven by usage telemetry (i.e. "engagement" metrics), rather than actually speaking to users about what they like and don't like. I suspect that companies sometimes believe they can automate their way into quality designs, by relying on analytics, rather than empathizing deeply with their customers.

> The initial call to deprecate this page was quite obviously driven by usage telemetry (i.e. "engagement" metrics)

yeah, they said that.

> rather than actually speaking to users about what they like and don't like

how do you suggest they talk to their users to do this? less than 1% of people invited to take a survey actually take the survey. however, when you say you want to remove something, those who use it make the effort to respond; at a far higher rate than would participate in a survey.

by saying they were going to remove the feature, they solicited better feedback than they would have gotten in any other way. if that was not intentional on some level, I would be surprised.

> I suspect that companies sometimes believe they can automate their way into quality designs, by relying on analytics, rather than empathizing deeply with their customers.

I suspect that some people will look for any excuse at all to share their opinion about the horrible new world with its horrible new ways as if only they and no one else knows what they are doing. I am not referring to you specifically, but your comment reminded me of just how often I see this "new ways to do things are always bad" sentiment and it is so wrong in so many ways.

Maybe add a page-improvement survey banner, where one option is to take the survey and another is to say "it does all that I need" and another is "I don't need it". Note on the banner that this is about the page not seeing much use.

very few people will even see that prompt. <1% of those that do will respond.

i encourage anyone who doesn’t believe me to spend time with UX researchers while they hold focus groups. it is absolutely terrifying what must be done to have a thin hope that 50% of your users will find something that they want to find.

i have always had very little difficulty noticing differences in pages i visit even semi-regularly, so it was very easy for me to dismiss those who don’t as “idiot users” and to write them off an unimportant. it turns out that people who don’t notice things that you put in a large font, bold, in the center of the screen are a very large portion of your users.

it’s frightening what people simply do not see.

I mean, how exactly do you argue with something like "0.5% of users have visited this page in the past 12 months, and 0.3% of users have only visited once"?

I don't know what their telemetry said, but generally it's something like this.

This is empirical enough to say decisively "most people do not care about this feature."

I love the github trending page, but I need only to visit it 10 times a year for it to fullfill its role. Yet I go to github every day.

So visits are not the proper metric to depict how useful it is. Just because I consult a doctor once a year doesn't make this appointment less useful than the pizza truck I go to every week.

This is bad logic. Almost everyone goes to the doctor in a year. And again, what is the importance of a doctor’s office. I once used a security tool that looked at the world in a very narrow sense. Despite being a general purpose tool, it would often flag really bad false-positives like “this security group allows all ingress!”, while applied to zero resources, while completely missing other, real, consequential issues. My point is, you have to pull your head up, look at the real world and the context of your decision, and use your judgement. These aren’t philosophical mind teasers.

It's not bad logic, it's an imperfect analogy which you are taking at face value instead of understanding the point that something can be useful even if infrequently used.

Wether or not it applies at this case is not determined, but it does underline the fact using only visit metrics for this particular page does not give the whole story.

Which is congruent with the reaction the announcement got.

To be clear, I don’t know what telemetry they are using, I only used percentage of users who visited as an example. It does sound like, of the small group of folks who use it, they do enjoy it.

Sounds like GitHub has to find a way to either expand that pool, or decrease the maintenance cost of the feature, or it’s possibly not worth it.

> This is bad logic. Almost everyone goes to the doctor in a year.

They do?

I'm not even making a point about developing countries and access to healthcare, I assume 'everyone' was implicitly 'everyone here' or in developed countries or whatever. But I would be incredibly surprised if that's true here in the UK. Is it an artefact of US health insurance that you can typically claim a 'check-up' per year (so why not even if not needed, you paid for it) or even that it's insisted on for the validity of the policy (ensure you don't have something expensive developing) or something?

> Is it an artefact of US health insurance that you can typically claim a 'check-up' per year (so why not even if not needed, you paid for it)

The point of a yearly checkup is because, at population scale, people can't know if it's needed or not. Tons of issues only present themselves subtly or not at all until it's too late, or just the fact that most people aren't knowledgeable enough to differentiate issues from normal body things. While less advised for younger people, as you get older it's advised to get a yearly blood check to make sure nothing is trending out of balance.

All that being said, I wouldn't be surprised if most people didn't get a yearly checkup in the US either, even thought it's fairly widespread common sense to do so.

Rather than just a “doctor”, think “brain surgeon”. Very few people have brain surgery every year, or ever. Only 0.5% of doctors are neurosurgeons. But brain surgery is an extremely critical service to co tinge providing.

Point is, the relative value of a feature is not necessarily correlated with its popularity.

if we follow that logic big red emergency stop buttons should be removed from machinery because of their low usage.

Sure. But good-sense would, in the case of emergency stop buttons, tell you that despite a lack of thought or usage, they are required for safe operation.

You can't make that conclusion with a webpage for content discovery.

You would need to model what kind of users you have on your website, And how they differ from one another.

Do they even that that feature even exist? I'd say a lot of users ignore most of the (social) features of GitHub beside "a public remote git repo I can download"

Understand your audience, then back it up with metrics. Otherwise we will easily exclude audience e.g. those who need accessibility support

0.5% of 100M is still a lot, right?

I was recently listening to GGG founder Chris Wilson and he was discussing how analytics can easily misfire if you do not apply them in a more conscious way. In his story, when reviewing logs someone saw that there was a big drop in engagement from quest A to quest B ( people stopped playing for a while between those two events ). What Wilson explained is that it is not exactly a surprise. Quest A effectively marks the end of a chapter and a lot of people stop at that point.

Chris's insight from that was and I am paraphrasing so hopefully I will get it right: 'It is ok to dive into data with a question you want answered in mind; we should be careful about stumbling onto data and assume the question as a result.'

Needless to say, it is a little different from, say, Amazon's creed so it is somewhat controversial for analysts.

That’s a good perspective from Chris. Did not expect to see this mentioned though with the current mutiny happening by their players.

But even if it not used by a lot of people, why remove it?

It's literally just webpage with a list of repos that have been starred the most.

It's not like it a complicated feature that needs a bunch of people to maintain lol.

From TFA:

> "The data pipeline that powers the page has historically proven problematic to maintain and so given the low usage, we decided to switch off the Trending page."


When you have a score (stars) people tend to start gaming it. Doubly so now

I've never seen the Trending page before, but it looks like another take on the the "Explore" page. I wonder, do people generally go to GitHub just to browse around? But if people find value in it, then who am I to say.

By the way, it's deprecating. Depreciation is something else.

The explore page is personalized, meaning it shows you repositories that are similar or related to ones you’ve already seen. It’s useful, but the trending page shows overall trends on GitHub and is a great way to expand your horizons and discover new things happening in the open source community.

"Explore" actually came _after_ "Trending".

It seemed, to me at least, that when "Explore" was originally created, it was intended to eventually replace the "Trending" page, using more personalised information. And the "Trending" page did languish, with more frequent breakages and less frequent updates, since they created the other.

> By the way, it's deprecating. Depreciation is something else.

This one is weird to me too.

If anything, Trending trending with this news has caused its appreciation

I like it for the more niche languages I follow such as rust/go. It's interesting to see where things are happening.

> for the more niche languages I follow such as rust/go

stares while writing Clojure

In an age where "no we are still correct and everyone who disagrees is just too myopic to see our grand design" has become the standard response to any backlash against corporate decisions, I have great respect for people who are able to simply say "we heard you, and we are reconsidering".

Luckily the headline for this post doesn’t contain any prejudicial language intended to disparage Microsoft for making this decision. Actually, wait a minute…

The most effective thing the team behind this feature did to drive adoption was to threaten to remove it. More people know about this feature now than ever would have through traditional marketing tactics.

Not sure how much this applies for others, but it's certainly true for me. Social media have tainted the word “trending” for me, so it's not something that I would normally click on when I see it.

[chuckling] I think yes; it seems they got the attention now.

Another commenter already pointed out this was going to be a https://en.wikipedia.org/wiki/Trial_balloon


This is a perfect example of why following Telemetry blindly is a shit approach to developing a product. Anyone who has used Trending knows it's probably the best feature on the site outside of actually working with a repository. In this case, the telemetry reflects not the usefulness of the page, but a total failing to promote its use.

I was actually shocked when I read the initial post because it showed a complete lack of awareness of how cornerstone this feature should be.

From reading the discussion about this issue here and on Reddit, it seems like most people weren’t aware that this page even existed - but the people who do know about it find it extremely useful. So instead of removing it, how about promoting it more?

As someone who didn’t know it existed. I concur.

It’s just a random link at the top of the page and I’ve never had any incentive to click it or look at it. But now that I see people find it useful it makes me want to check it out daily for a while to see what shows up on there.

I think the only reason I know about it was that I believe trending was the old explore page.

Most of the folks have already mentioned on the thread; it's a pretty cool place to find something that one does not know even exists.

Quote from the thread[0]:

''' Hey folks - Martin from GitHub here.

Thanks for all the feedback on the Trending page. Given the feedback here, it’s clear we need to look again at the plan to deprecate this. The team is going to re-evaluate and see if we can come up with some other options.



I remember when this feature was first launched, GitHub got lot of heat from devs saying it's turning like a social media site and they didn't like it. But, apparently many devs liked this feature as well (including me). The toughest thing about product decisions is keeping different set of users happy :)

Ha! I'm on github nearly every day, and I never noticed the "trending" page. Just took a look at it now to see what's there.

In general, I don't spend nearly enough time exploring things.

I am incredibly glad they are keeping it around for now -- for now being the key word here. From the way they worded it, doesn't seem like they were prepared for the feedback they received, in terms of (paraphrased) "the few people who used the page seem to love it". How do they not know such an important metric? Shouldn't they be able to measure the number of recurring users? Shouldn't that metric tell them that instead of shutting it down they may just need to promote it better?

I think it should be obvious that they measure lots of things and that the data says the page isn't used very much. This is a case of the data not telling you the full story. That happens sometimes. Data needs to be interpreted, and sometimes people get it wrong. There's probably a lesson in there somewhere.

However... This is tech, and the way engineering teams in tech work these days means it's probably something as simple as this feature is considered low impact and not "career enhancing", so no one was willing to take on maintaining it or promoting it. The only option was to kill it. When someone saw how passionate users are about it they changed their mind.

Lots of devs use ad blockers that block telemetry

I'd expect such telemetry to be entirely server-side

People new to GitHub may want to look at the Trending / Explore pages to get a feel for the site. Those who are frequent users may skip that in the normal course of their business but drop in once in a while, if they are curious.

It is okay. At least in the segment of GH users who cared about the feature, there was a significant enough outcry from GH's perspective that they chose to rollback their decision.

People new to GitHub want to create a repository and do a git push, nothing else.

Ooh guessing game! Here's my entry. People new to GitHub arrive by searching for <software>. They read the readme.md and then follow the installation instructions.

I knew someone would comment this, but I wouldn't consider those as "GitHub users", because they are not registering, and are not interacting with anything on GitHub. They only care about the specific repository they are viewing, not the platform.

They are incompetent honestly. If you were to discontinue this page you would at least talk to a few power users of the page and know their mind before taking the decion. They either didn't do it or did it. If they did it but backtrack now it show they don't have much confidence in their decisions, and if they didn't it, what are their pm payed for? That is pm 101

> at least talk to a few

This is more like how to get unrepresentative sample 101.

Maybe they did and concluded it's fine to remove only to find out that general opinion is different.

> They are incompetent honestly

Do you have to insult them? People make mistakes all the time, taking decisions is hard. It's clear they didn't expect removing this page to be a big deal.

The fact they acknowledged the issue makes them the opposite of incompetent IMO (inincompetent!)

lol I can't believe the bitchers and moaners here (and I assume elsewhere) got them to change their minds.

The proper usage of github is: go to the main repo main page to get the clone URL; go to the issues page to file an issue if the projects uses them; go to the PR page to make a PR if the project uses them. Every other page may as well not exist.


90% of what's in that page is information you literally do not want when you come visit their website

People working at Github have no idea what their website is all about

The "trending" culture is a death sentence to our civilization

Commend Github for listening to feedback.

I don't get it if nobody used it, why keep it? Don't the stats tell the truth here?

The stats tell whatever truth you want them to. They're used to rationalize decisions more often than inform them.

I wonder how much of that page is the result of "voting" rings/bots.

What sort of telemetry are they using for this, I wonder?

Applications are open for YC Winter 2024

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact