Hacker News Reader

Comment by throwaway894345 | Hacker News Reader

Sure, GH lags SO considerably on the size of the knowledge base, but I don't think that's very important since, due to the rapid pace of tech, content rapidly loses value with age (consider all of the jquery or flash content, or content pertaining to old versions of libraries or frameworks).

What matters more is the rate at which content is generated, and I think GH has a pretty competitive story considering that people naturally want to ask the project maintainers/community questions directly (as evidenced by all of the GitHub Issues which are formulated as questions) and considering GitHub's significant share of those maintainers/communities. I posit that new content will increasingly appear on GH Discussions.

Please don't mistake this for some exaggerated "GH is going to kill SO" argument--SO will do fine, but there is a compelling story for why MS wouldn't spend $1bn on SO.

View on HN

> Sure, GH lags SO considerably on the size of the knowledge base, but I don't think that's very important since, due to the rapid pace of tech, content rapidly loses value with age (consider all of the jquery or flash content, or content pertaining to old versions of libraries or frameworks).

This is a weakness of SO and really the whole knowledge base business model (Q&A or not). You think you're building up this network-effect moat. "We have the best question-askers, answer-givers and answered questions. So everyone comes here!"

But, knowledge goes stale. "How do you X in JS?" is different than it was 10 years ago. People don't want a historical archive of how it was 10 years ago, they want to know today!

Then, like you say, GitHub brings in a whole new weird angle. Why ask on some other site when I have a spot where the maintainers and users hang out? Since my loyalty to SO is approximately zero, I'm just as likely to click on the Google result that takes me to GitHub if it looks more promising.

I think it's perhaps a little clearer to frame it this way:

SO has a network effect moat, not a corpus-size moat (I don't know that SO was ever thinking the size of its corpus would stave off competitors--they probably were correctly assuming that their user-base was their moat). The network-effect moat keeps the rate-of-change of the corpus higher than competitors, which is where the value is (not the size of the corpus). This moat has held up really well apart from (maybe) GH Discussions, which seems (to me) likely to succeed because GH has a similarly large network of experts.

We agree on the corpus-size not being a moat, and maybe SO does themselves.

More directly, I question StackOverflow's network effect. Compare it to Craigslist, IMO the best example of network effects ever. It is their sheer size that has entrenched them. They're also not vulnerable to groups moving to another platform (rise and fall of everything from AIM to Snapchat).

I don't actually search StackOverflow. I end up there through Google. If Google starts pointing me somewhere else, or the search previews start looking better for somewhere else, I'm going there. It's not like if I try to sell a single bag of concrete on anywhere but craigslist, which would end in failure.

I'm sure in the M&A docs there was some disclosure like "75% of our traffic comes through Google. If Google started directing people to other sites, that would have a material impact on our business."

SO has a SEO moat.

As soon as the answers I need start appearing in some other site, I will use the new site.

> Since my loyalty to SO is approximately zero

That's you though. It's not about loyalty, it's about Stackoverflow almost becoming a verb like Google. Many many people are used to using it, it's brand is very powerful. I don't see some competing tool stealing lots of mind share. Btw the thing about talking to maintainers directly is already done on small communities on discuss pages (ElixirForum for instance), on Discord, IRC and many other channels. None of it actually hurt SO.

Brand is not a moat.

How do you define a moat? For me it's anything that gives a company an advantage over it's competitors. It can be patents, tech, but also brands. Take Coca Cola for example. Now obviously it has a moat right? Where is this moat coming from then? If you think about it it's mostly the brand, not anything to do with taste. Same can be said about different beers or about Nike, Levi's, Apple (arguably) etc etc.

Good question. It might be a difference in degree rather than a difference in kind. Consider the social media space: if you want to build a Twitter clone, you could do so in a pretty short amount of time with a pretty small team. You could buy ads to get everyone to know your product's name. Those ads will bring in enterprising early adopters, but no matter your ad spend or the quality of your platform (you could even include an edit button for pete's sake--and for free at that!), you're very, very unlikely to unseat Twitter. The reason is that Twitter's product isn't its platform or even its brand--the product is the social network i.e., the network of users and the interactions between them. That's moat.

> If you think about it it's mostly the brand, not anything to do with taste. Same can be said about different beers or about Nike, Levi's, Apple (arguably) etc etc.

Brand is king in fashion (and to a lesser extent, low margin consumer products like cola or cereal) because fashion is largely about signaling status. I don't think this effect extrapolates to SO.

I think Twitter is mostly about momentum, everybody is there and people dont want to start over again. A competitor will have the critical problem of having zero users. I am not sure the connections between users matter that much. Not all brands is about status. How can cereal be about status for instance? A lot of the crap we buy and services we use is just habit and acquired behavior. Sometimes status signaling is part of it and sometimes it isnt. Is capncrunch about status lol?

TomTom and BlackBerry had great brands, marketing and great software stacks. I still correct the google maps voice when it tells me to take the third exit on the roundabout. It’s “roundabound”. The people who ran those companies didn’t see anybody stealing their businesses either. Yet they are well and truly gone.

https://youtu.be/ATulnxruvhQ

Not sure TomTom had as dominant a brand for navigation as Stackoverflow has for Q&A. Stackoverflow is pretty much alone in this space still which is pretty amazing when you think about it.

Quora?

Ah no I meant technical/developer Q&A of course

> I still correct the google maps voice when it tells me to take the third exit on the roundabout. It’s “roundabound”.

If it's saying "roundabout," I hope you're not correcting it to "roundabound," because "roundabout" is the correct word.

https://en.wikipedia.org/wiki/Roundabout

Yeah what? Maybe it’s some local dialect.

> But, knowledge goes stale. "How do you X in JS?" is different than it was 10 years ago. People don't want a historical archive of how it was 10 years ago, they want to know today!

I want both! With commentary!

And SO is particularly hostile to this, with their "duplicated question" giving canonical status to obsolete answers.

Good point. When I wrote that comment I was imagining the value of something like Datomic, which could let you see the result of a query at any point in time. In 2014, how did someone answer this question?

GH Discussions is great when the question is closely tied to a particular repo. I don't see how it competes with SO for things like design patterns, algorithms, programming language usage, etc.

I wonder how much of SO fits in the "particular repo" category? Not so much for my interests, but I imagine the chunk for popular web frameworks is huge.

Anecdata incoming: a few years ago when I was doing stuff with c# it seemed like everything I needed was already on SO answered by some dude with like a million kudos or whatever.

Recently I need some python + mariadb answers and most of the top ones are out-of-date enough that I can't even use them (deprecated packages). I was a bit surprised TBH.

Also TIL GitHub Discussions is a thing.

Yeah, GHD has not been well-marketed, but I've used it a couple of times and it has been very useful. I have similar experiences where SO questions can be pretty dodgy based on age--for example, for Rust I rarely bother looking at answers from 2016 because so much has changed since then. Some technologies are more stable than others.

Usually a highly voted answer that is no longer correct will be updated through time. I don't have any data to support this claim but it's what I see happening.

Probably, but that furthers my point: that "update" is new content. The original content is not especially valuable; the new content--the user activity--is where the value is.

“Average technology has 3 answers a week” factoid actualy just statistical error. average technology has 0 answers a week. Jon Skeet, who lives in cave & answers over 10,000 each day, is an outlier adn should not have been counted

I'd say a lot of that stuff increases with value to an individual over time, but has less demand.

Some of those SO discussions are the only references to ancient legacy code that is useful, when asking any questions about it today would not get any answers.

So, depending on how we calculate value, if we ignore demand, it can be more valuable.

Is it just Stack Overflow or Stack Exchange over all? If the latter, that's a lot more content that doesn't expire that quickly.

The latter. The initial wordings described multiple q and a sites in the article.

Right, but MS probably isn't interested in that content for its software dev portfolio.

Reply on news.ycombinator.com