Results from being #1 on Hacker News for 6 hours | Hacker News Reader

Results from being #1 on Hacker News for 6 hours

codeamigo.dev

139 pointsplondon51439 comments

> Amazingly, AWS billing stayed very low/consistent with other non-active days

The way they are showing CPU I'm guessing they are using a VPS, :. the only variable cost is probably egress, which is often incredibly cheap (compared to a non trivial server load).

Or, in other words, yet another time when you don't need k8s!

I reached the front page of HN too (though not #1). My rough stats were:

20-30,000 new visitors

Most were from USA, with europe and India, China behind.

Mobile users were 70-80%

I've been on top of HN and Reddit programming before as well. Though you get a boost, the traffic isnt very sticky. From those 30,000 visitors, I got maybe 20-30 email signups. And the traffic dropped in 1-2 days.

But there are long term benefits, as the link was then shared by other people on Linkedin, Chinese new sites etc, leading to indirectly new visitors over time.

Fascinating. I thought that HN was more like a 100K community (of monthly active users). I guess my ballpark is correct? I now think it's slightly less.

I don't click on every #1 link.

I probably click the #1 link less than 10% of the time and idunno there's only like a 1 in 3 chance I check HN when any given story is on the front page, so from their 29k views now I'm thinking HN is 1M active users. Which of us is way off?

Some guestimating: I clicked on 3 of the current top 10 posts, so if that's a representative click-through rate then that would put the HN community at ~100k daily active users (which probably means more like 1M monthly). That's assuming the post stayed in the top 10 for 24 hours, or at least long enough for visitors from every time zone to notice it.

A 6 hour window is less than 1% of a month. If you think HN has 100K MAUs then you should expect way less than 28,000 people to hit HN during that 6 hour window to see the link in the #1 spot. Even if the article was on the front page for a full day you wouldn't expect 28,000 clicks because fewer than that would be hitting HN on that particular day. Traffic isn't distributed linearly across the month but still.

Even then, I don't think this is a representative snapshot of HN users anyway. The average time on the site was 37 seconds. That's enough to work out what it's about and decide it's not interesting (to that person). As Codeamigo is a coding school, and HN has a lot of users who can already code, that's not surprising. It's likely that far more people read the comments first and didn't bother clicking after understanding what it is.

More importantly than all of that though, is that the link wasn't about Codeamigo. It was about starting it after being rejected by Codacademy. The interesting part of the narrative was about coming back from rejection, not the product itself. The fact 28,000 people clicked at all is amazing.

This would only be correct if stickiness (the DAU/MAU ratio) were around 3%, i.e. if each MAU only visited once per month.

You know what would be nice? HN stats! Active users (daily, monthly, etc.), how many posts a user checks on avg., how many comments per user, how many submissions per user, top commenters, top etc. etc. etc., breakdown per OS more etc. etc. etc. :)

@HN, what do you think?

I think it'd be hard to back into that number only based on the analytics of one top-page link. I had almost a quarter million hits from a #7 post in January.

Here are more details:

https://forum.photostructure.com/t/front-page-of-hacker-news...

From Dang [1] in 2020.

>These days around 5.5M page views daily and something like 5M unique readers a month, depending on how you try to count them.

And that is excluding API access.

Generally speaking I think it depends on the topic and time of day it is on HN front page. Since HN is mostly US specific so it starts at around GMT-4 during working hours. Yes. People surf HN more during works, that is why you see lower submission on weekends.

I am also suspicious of decline in page view on HN. Compared to pre 2020. At least the upvote number on front page and comments seems to be trending downwards.

And yes, I do wish HN could have more stats available

[1] https://news.ycombinator.com/item?id=23808787

HN has been growing at the same roughly-linear rate for many years, including this year and last year.

I haven't checked, but I'd be surprised if upvote numbers are going down.

I personally very rarely actually go check out what is linked on HN, due to mostly using it on internet connections with very restricted speeds.

I also know a fair few people in my social circle that do the same, HN is the goto "I need good content that doesn't zap 100mb of data in 3 minutes" place

I was #1 for a few hours a year ago, from last St. Paddy’s Day. It was a nice ego boost, but it only lasted for a day or so, and then it was back to being a regular schlub.

Anyone interested in a secondhand cape? I never even got a chance to wear it.

I've done it a few times, which shocked me. Each time it meant about 20-40k unique views and some solid discussion. That's honestly my favorite thing when it happens, seeing something I wrote about discussed in depth on here by people who know what they are talking about.

The only real lasting benefit was a steady stream of traffic to those articles years later.

Reality Driven Development: https://news.ycombinator.com/item?id=17154355

Comparing Elixir and Go: https://news.ycombinator.com/item?id=13497505

Docker is the Heroku Killer: https://news.ycombinator.com/item?id=8371249

No Such Thing as "Real" Programming: https://news.ycombinator.com/item?id=11045526

If this post gets #1 I expect to see results titled "Results from being #1 on HN about being #1 on HN"

Ask and ye shall receive

Can you post OS stats as well please, if you have them?

From my own posts which made it to the front-page, I remember that non-mobile users were predominantly on Macs, with Linux far behind and Windows being tiny. I don't remember the exact numbers though; and I'd also love to hear the Android/iOS breakup.

Operating system Visitors iOS 8k (28%) Mac 7.9k (28%) Android 7.2k (26%) Windows 3.5k (13%) GNU/Linux 897 (3%) Ubuntu 376 (1%) Chrome OS 94 (0%) Fedora 68 (0%) FreeBSD 4 (0%)

Thank you.

> Investigating the Vercel response headers, we can see that the X-Vercel-Cache Header is 'MISS'

Lee from Vercel here. SSR + caching is extremely powerful, but it's more difficult to get right :) We're working on providing more tools to help developers understand if they're not adding the correct caching headers, but in the meantime, tools like Incremental Static Regeneration help you have a static output while still having dynamic data. It's not a perfect use case for everything, but it might be something to consider. Regardless, glad you got the caching headers correct - let me know if you have any other questions or feedback!

Interesting! I only consume HN via RSS feed. That means: chronological order + I have no idea what is #1 at any given time. Always wonder why some posts get tons of comments and others almost none at all. Micro echo chamber? Does the time of posting affect the impact? I wonder how many people have the same experience.

I once linked a video on my YouTube channel in a comment which was a reply to the top comment on a thread.

My YouTube channel gets hardly any views, but that one day I got perhaps 100-200 hundred views.

But the knock-on effect was great (although small), YT finally started recommending my video to others, and I was averaging between 1-3 views a day. It’s tapered off now to a handful of views a week.

Possibly useful data for other startups: SimulaVR (https://www.simulavr.com) made the front page a few weeks ago. This led to ~2.5K new signups for our Linux VR headset waitlist.

If you can handle a HN surge with a low end droplet on Digital Ocean...... this ought to give some sense for the of us using Kubernetes - it is not really needed _most_of the time!

I don't understand why this blog post has invited non-sequiturs about Kubernetes. Autoscaling is only one small part of what Kubernetes provides.

Well, I for one have never heard anyone "evangelize" other use cases. Can you name some advantages, or point me to an article? Because I did hear about the set up difficulties...

I thought it was all about easing container deployment and orchestration (not that I know anything about it).

resilience to dropping servers is one. (sort of hand-in-hand with auto-scaling, both imply dynamic sizing of server pool)

I can only guess that small sites frequently experiencing the hug of death when they get #1 on HN leads some people to believe you need some insane setup, when in reality it's more like this case: it works just fine on a smallish setup as soon as you remember to enable caching.

it usually amazes me to see sites going down because of HN traffic because it is really not that much, in fact barely noticeable in a 10-year old server.

I run Kubernetes on my low-end Digital Ocean droplet.

Do the raw numbers take into account adblockers?

Hmm good point (it does not). Codeamigo uses Plausible, which is blocked by adblockers.

I can't tell, is that bounce rate mentioned in the article good or bad?

i ve had posts in the frontpage in the past. It's not worth it for the community, it is negative and stuck in a groupthink. It usually contributes zero or less than that. Still, you get a lot of link juice because a lot of other sites copy the link.

It's been a pretty positive experience for Codeamigo. People have been reaching out to help out, submit bug reports, and suggest interesting features.

is the time zone in the plots CET?

EST

Reply on news.ycombinator.com