Show HN: Inflation-adjusted Hacker News

https://instruments.digital/inflation-adjusted-hn/

by Aaronmacaron • 5 years ago

437 119 5 years ago

wheybags• 5 years ago

The extension is available for Firefox and Chrome. Unfortunately the extension is only available as a direct download for Chrome because I didn't bother paying Google money to publish this extension. The Firefox extension is available in the official Add-On store.

This is so sad... I knew they required a paid account to submit to google play, didn't realise that applied to the browser extension repo as well. Oh well, just another reason to stay on Firefox :p

rnotaro• 5 years ago

It's a single 5$ registration fee per account. Nothing huge.

Once you paid the fee, you can publish as many extensions as you want.

https://developer.chrome.com/docs/webstore/register/

bredren• 5 years ago

This is probably a very light anti-spam measure?

swiley• 5 years ago

Oof you have to pay Google to publish extensions (and they still can't manage the spam?) That's some brain damage.

berniemadoff69• 5 years ago

firefox now requires developers to 'sign' their extension with mozilla, and has disabled the ability to side load extensions manually for more than just temporary use, unless you are using some special developer version of firefox. chromium still allows users to side load their extension. it's ironic that google would allow more freedom to the user than mozilla.

jlokier2• 5 years ago

I rarely use Chrome. About once a month to make invoices (Firefox PDF output isn't correct), or rarely for some site that's not working in Firefox or Safari, just Chrome.

Even with that low usage I had to side load an extension recently to add some functionality. Thank goodness it's still possible.

luke2m• 5 years ago

Maybe that's good to reduce spam submissions.

andrewljohnson• 5 years ago

Seems like a good way to prevent a lot of bad actors spamming. Google has a lot of problems with people automating free signups for computing and distribution resources... if you give someone any amount of free computing power, abusers will come.

greesil• 5 years ago

The 500 point karma limit should probably be inflation adjusted as well.

gnicholas2• 5 years ago

Karma doesn't accumulate linearly. I don't know exactly how it works, but certainly for articles I've posted it's true. It might get 300 points but your karma only increases by 200. I don't know if this non-linearity has changed over time, but it could certainly be used to achieve this effect.

And to be clear for others who are wondering what happens at 500 points, it's a threshold for being able to downvote, not a limit per se.

zepearl2• 5 years ago

But then probably it should rather be the points that get assigned that would have to be adjusted rather than the thresholds?

Reason: if I e.g. got 1 point 2 years ago, that point had at that time more value (less users than today) than the 1 point that I might get today? (therefore today I should maybe get just 0.5 points per upvote instead of the 1 point that I got 2 years ago, depending on the difference of active users throughout time)

The other approach of just raising the thresholds would on the other hand impact all my points independently from "when" I got them, which sounds wrong to me.

midasuni• 5 years ago

Many years ago I learnt the difference between less and fewer, and now I can’t read or hear someone saying “less than” when it ‘should be’ ”fewer than”

It’s like Roko’s Basilisk. Or The Game (which I just lost), living rent free in my mind, offering nothing of value, yet immune from bleaching - indeed alcohol if anything heightens it’s power (and causing rambling asides, I think ki should go back to torturing myself with Lithuania’s Eurovision entry)

waiseristy6• 5 years ago

I think that the downvote mechanism should just be removed and the flag cap moved to 500. It's just become a mechanism to remove controversial topics from conversation. HN used to have some very off-the-wall libertarian voices which incited some interesting conversations. Now everything just gets downvoted into oblivion unless you are Mainstream Approved

Another interesting point as I sit here and watch this comment literally 'fade' into the echo chamber. Why is it that HN signals immediately when comments go below the 1 point threshold, yet hides upvotes? It seems like HN encourages the bandwagon effect on downward pressure only

Taek2• 5 years ago

I do think the downvote button contributes considerably to shutting down things like pun threads and low effort comedic jabs. It's always sad to see someone who is making a good faith effort to have an interesting conversation get drowned in downvotes, but on the net I think the downvote does more good than harm.

nanidin2• 5 years ago

I think that the downvote mechanism should just be removed

Why? I think this approach makes an assumption that everyone is upvoting everything that they would not downvote. I upvote things I find interesting or insightful, and I do my part in terms of moderation by downvoting comments that don't have a good fit for HN.

What I see on Facebook and Twitter is that the inability for the hivemind to have some kind of downward pressure mechanism on comments encourages/enables polarizing comments by vocal minorities to rise to the top. I see this frequently in the comments on the posts of local political figures. The top comments shown are almost always controversial and unhelpful/unproductive - things like whataboutism, blatantly racist remarks, etc. Eventually I got tired of constantly catching myself up in arms over FB comments and stopped participating.

The only way I see to counteract this as a rank-and-file user is to upvote/like every single comment except for the ones that are hurting actual discourse. IMO this will drive away people that you ultimately want to be participating on HN.

imtringued2• 5 years ago

I personally think votes should only affect the order of comments. I am interested in reading down voted comments just to avoid the echo chamber effect.

miguelrochefort• 5 years ago

I thought this would be about compensation figures.

It’s about HN points.

neogodless• 5 years ago

I've accumulated a lot of HN points, and at no point did I discover some way to turn that into compensation!

krapp2• 5 years ago

The true karma was the friends we made along the way.

snowwrestler• 5 years ago

I once saw a comment here that some people put their HN usernames on their resumes.

I was shocked; seems like that is more likely to be taken as a negative signal than a plus. “Here, look how good I am at wasting time on this web forum!” :-)

ingve• 5 years ago

It's not quite compensation per se, but rsync.net will give you a very nice discount just for being a HN user. (I don't think they check how many points you have accumulated though.)

eatonphil• 5 years ago

There's a similar issue on reddit. For a growing sub (which most interesting subs are) sorting by top of all time and top of this year is the same thing. I wish there were a way to sort by top by year on a sub (and HN too of course).

srathi• 5 years ago

https://redditsearch.io may help a bit with these types of queries. Bummer that there is no native way to do so.

ALittleLight• 5 years ago

You'd think this is the kind of issue reddit could solve instead of figuring out how to spam you with emails faster or creating snoo avatars.

cratermoon• 5 years ago

Like most social networks, history is largely irrelevant to how they make money. They need to keep you clicking on the new, the ephemeral, and the unusual, serving up ads and collecting information about your interests, hobbies, habits, and biases to refine the ad targeting.

After a year or two "most popular" no longer means much for those sites.

dheera• 5 years ago

Another effect to consider is that given that the number of "slots" on the front page hasn't changed, the time you have to rack up threshold upvotes to hit the front page will have reduced. New posts fall off the "new" page much quicker.

Might be interesting to plot the number of new posts per day over the years as well, if that data is accessible. I would guess it would inflate at the same rate.

JonathanMerklin• 5 years ago

For HN, you should know about: https://hn.algolia.com/?dateRange=custom&page=0&prefix=false...

DoreenMichele2• 5 years ago

The Lists on HN are also useful.

https://news.ycombinator.com/lists

crazygringo• 5 years ago

I'm curious how closely this would match overall HN traffic -- if only very loosely, or if almost pixel-for-pixel.

Only @dang could tell us!

(It's not at all obvious to me that as HN has grown more popular, whether the proportion of voting to non-voting users would have changed as well.)

etaioinshrdlu• 5 years ago

I'd love to see a list of top stories of all time - where the ranking was inflation adjusted!

ipaddr• 5 years ago

A list per year works and would be easy to use to gage the inflation if you wanted an all time list.

paulpauper• 5 years ago

The number of points a story has accumulated on Hacker News is a good indicator for how relevant and overall how good the story is.

Sorta. I think it is also demonstrative how much effort there is or how popular the subject is, but this does not mean it is good. If I could change things, it would be that some stories do not stay on the front page so long. Stories about apple news tend to get a lot of points but I don't think they are the most interesting, at least in my opinion. I think there are too many science stories, from maybe 3 science sources(quanta magazine, nautilus,and one other).

the_only_law2• 5 years ago

I end up having to block a lot of the crap that reaches the front page because it otherwise floods the comments link with low quality bickering when I’m trying to hunt for interesting threads.

Taek2• 5 years ago

Curious if you have a list of topics that uniquely/surprisingly stand out as high or low quality

pbhjpbhj2• 5 years ago

I'm looking forward to having an AI that blocks all the "this animal I keep locked up is cute" posts, they do nothing for me; I just don't want to see them.

A Slashdot style -30% of points for posts tagged 'cute' would probably do it too.

dang3• 5 years ago

What are some examples of crap and how do you go about blocking them?

p1necone2• 5 years ago

I upvote based on how interesting/valuable I think the discussion will be, not necessarily the quality of the article itself.

derbOac• 5 years ago

I also have this hunch that submission timing plays a role, not only on HN but other places where there's some time factor involved in the moderation system, explicitly or implicitly. My guess is you could examine this somehow quantitatively, although it might be a little idiosyncratic to specific locations.

gnicholas2• 5 years ago

This is cool, and it would be fun to see top stories by year, and top Show HNs by year.

I posted a Show HN [1] when I was new-ish here, and I didn't realize how unusually successful it had been until I saw we were getting web traffic from a site (now defunct) that listed the top 10 Show HN posts of all time.

I've periodically used the HN search function to see what the top HN posts have been, but with the point inflation it's become difficult to make meaningful comparisons across time.

1: https://news.ycombinator.com/item?id=6335784

ammar_x• 5 years ago

Nice and interesting analysis. Thank you for sharing.

I now think HN and Reddit and others should consider inflation by default when you search for top posts of all time for example.

pbhjpbhj2• 5 years ago

Top contributor lists should probably have a date-based window [option?], so they show scores for the last 12 months, say.

crackercrews• 5 years ago

It's basically the average point count of the top 100 stories of each day.

How sensitive are the results to changing 100? Using a smaller number would be more accurate for the top posts.

You could also use an algorithm that uses different calculations based on the popularity of the original post.

soheil• 5 years ago

Not only this but also the number of votes from users who were active years ago is also a very interesting metric to see. No doubt HN audience has changed over the years and votes from original users are diluted by votes from new comers. It won't be easy to do but one could aggregate "favorites" from all users on a daily basis and only count the votes from users with an old registered date.

abnry• 5 years ago

The shape of the curve is fascinating. You see exponential growth in the early part, which makes sense, but then it sort of converges to a bumpy linear function. Network effects would suggest power law growth, which linear fits into, but this just seems too linear. Normal currency inflation is usually measured as a percentage per year, which implies exponential growth. What is the current %points change per month, for instance?

jkuria• 5 years ago

The other consideration is that the traffic (and submissions) have grown likely 50x plus but the front page still only has 30 slots, and so the 'needle eye' is a lot smaller.

One could argue that a story that makes it to the front page today is much higher quality than one that did so years before.

You get more for 100 points, just like in economic inflation terms, you get more for a consumer electronic that has stayed roughly the same price but quality and features have increased considerably.

bastijn• 5 years ago

Maybe. Though one should then also account for the likely case that early adopters were more niche and deeper into the content, being part of an early community. When communities grow it attracts all sorts of people which may decrease the quality of content but because there is more a few gems will still occur. Because they would be hidden behind the "crappy" content they need to be given more points to arise the mass. A prime example is Stack Overflow which almost died due to its own fame (for the early adopters).

caslon• 5 years ago

The front page has definitely gotten worse over time. More people doesn't actually imply better judgement in story quality. The moderators have admitted that their goal is to slow the inevitable decline in the quality of the site that comes with a bigger audience; they have no illusions that the site would somehow manage to break the only replicable rule of the internet, that any site will get worse with a wider audience.

Retric• 5 years ago

It fluctuates but I think overall it’s been very consistent over time and I joined in 2007. Some of the low points where Bitcoin article spam, and various political issues. But, I still see plenty of of stuff from time to time including a very long tendency for random Wikipedia pages to make the front page.

Edit yep there was one: https://en.wikipedia.org/wiki/Initial-stress-derived_noun

sbagel• 5 years ago

Agree with the 'needle eye' being smaller but

One could argue that a story that makes it to the front page today is much higher quality

This assumes that quality of content is the driving factor in success on Hn or any social network which isn't the case. Timing, attention grabbing title and an engaging subject are arguably more important.

swyx• 5 years ago

fyi dang has disclosed traffic details on a semi frequent basis - it is currently at 6m requests a day, up from 4m requests a day in 2018.

https://twitter.com/swyx/status/1371586759578984450

lifeisstillgood2• 5 years ago

Brilliant work. And it makes me think - we produce tools, that are part of two phase spaces - the idea and the knowledge of how to create it. Firstly I simply would not have thought of it, "like button inflation" is a pretty off the wall concept. On top of which I don't know where to get an HN data feed or how to make a chrome extension etc. Yesterday I was on a call with a collegue and the conversation turned to a useful feature we did not have (internal use) - and 30 minutes after the idea we had coded it up - because we had the knowledge but only lacked the idea till that moment.

Without that knowledge even the best idea moves the process from "yeah let's do it" to "I should schedule some time and make a project plan and ..."

The difference is important somehow. And most of our working lives would be better if we stopped with the project plans and just built useful stuff.

Edit: To try and clarify - exploring a new and uncharted phase space is exploration - it cannot be planned in advance, or driven in a particular direction. Sun-Tzu teaches that Climate, Landscape and doctrine all affect our success as much or more than mission and leadership - so we should stop trying to control and actually find the flow.

I think we search for product-market fit, we do not decide product market fit.

One tool I realise i have been trying to build on and off is a post-hoc reporting tool that makes it look like my exploratory development style was actually planned (it's always something like, I am going to explore this inferring hill over there, and just make sure the roadmap, docs and timesheets are up to date.) This way I don't get people bothering me so much. It works better than grumpy.

lifeisstillgood2• 5 years ago

Side note:https://podcasts.apple.com/gb/podcast/more-or-less-behind-th...

That podcast talks about building recovery medical trial - it is a near perfect example of the "next door in phase space" thing i talk about

forgotpwd162• 5 years ago

All 1 point show as 2 points. Shouldn't those be 1 point regardless of date since it's the base and it means some specific post had no upvotes?

hawski• 5 years ago

Very nice realization of the idea. The extension is a sweet perk.

I was thinking about something like this some time ago. I also wonder how it compares with comment count per post and score/comments ratio in time.

the_cat_kittles• 5 years ago

this is a wonderful topic that i have always kinda wondered about informally. a slightly adjacent question ive had is how much does the rank of a post feeback into votes. my assumption is that the higher the ranking, the higher the absolute number of votes the ranking itself (independent from the actual article) generates. coupling some kind of way of compensating for that with your historical inflation adjustment would give you a fairly uniform metric for gauging public reception to submissions i think.

DoreenMichele2• 5 years ago

I bet you see a lot more comments on average on popular articles as well.

tomjen3• 5 years ago

This happens on most sites, including reddit, but the opposite happened on stack overflow. I started posting there when I was in uni, but mostly stopped when I got a job.

Yet because the traffic has grown, those comments keep (or kept) accumulating karma. More interesting, because I answered some questions that became highly popular, I get a much larger share of the karma than others who have had to spend more time researching.

AlchemistCamp• 5 years ago

Interesting chart. What happened in April 2012?

shoto_io• 5 years ago

The two dips in 2013 and 2014 are interesting. It's definitely not a straight line up.

shaicoleman• 5 years ago

FYI: If you read Hacker News via RSS, check out https://hnrss.org/ , which allows you to filter the feed by points/comments.

e.g.

https://hnrss.org/frontpage?points=100&comments=25

Liskni_si• 5 years ago

Would be nice if hnrss filtered by inflation adjusted points… :-)

Scene_Cast2• 5 years ago

I'd be interested to see a "top of all time by adjusted score" list of posts.

webmaven• 5 years ago

It would be interesting to see a 'best' page based on inflation-adjusted scores:

https://news.ycombinator.com/best

That would require indexing the inflation adjusted scores of every post (or at least, every post that has ever made it to the front page, or maybe just every day's #1 if you assume that no day's #2 and below could ever outrank any other day's #1) rather than adjusting the score of posts on the current page, which is more than can reasonably be expected from an extension, so I'm not actually making a feature request... unless you feel like it.

brntsllvn• 5 years ago

Love the web app, but my browser extension space is so limited I'm not sure I want to add another with such marginal (however interesting!) value. Thank you for posting.

MattRix• 5 years ago

Seems like it would be interesting to see an inflation-adjusted top 100 stories of all time.

ineedasername5• 5 years ago

Pretty soon 100 karma won't even buy you a gallon of milk, much less the small 2-bedroom colonial home it used to back in '08 or '09. (Not counting east or west coasts... Hard to find anything under 300 karma there.)

At least it's still harder currency than FB likes or Twitter retweets. Those internet points might as well be $5 billion Zimbabwe notes.

nolok2• 5 years ago

The really funny part being of course that there is an actual business out there out of fb likes, instagram followers and retweets

MichaelMoser1234• 5 years ago

https://blog.saleslayer.com/selling-on-tiktok#influencers They tell businesses to partner with tiktok influencers for direct advertising, so that some people are actually able to turn internet points into money (by analogy: maybe there is a point for businesses to partner with any one on this list: https://news.ycombinator.com/leaders ?) Also a lot of stuff on youtube contains some form of direct advertising as part of the video.

stjohnswarts• 5 years ago

what's the difference of making money off of fb, insta, and twtr rather than off radio or tv ads?

reputet• 5 years ago

I guess we can expect the karma devaluation.

ineedasername5• 5 years ago

It's really a strong case in point for why karma needs to be mined in a PoW blockchain. I think if HN did an Initial Karma Offering (IKO) there would be plenty of takers. Airdrop some to Elon at the same time and we'll be all set.

datameta• 5 years ago

I have noticed the opposite happen on Dwitter.net (a visual creative coding community) - I think it takes more thought, time, and effort to come up with new code tricks (or use existing ones in a new interesting way). Similarly, certain visual motifs have been explored and seen by those who have been on the site for a while. There is still a steady influx of people so I think it is the nature of lower-hanging fruit becoming less impressive to the average dweet enthusiant.

dools• 5 years ago

Pretty soon we will have to start borrowing karma from China and leaving a debt for our children. The only solution is to create a deflationary blockchain based karma solution that makes fewer votes get recorded over time based on increasingly difficult proof of work.

nprateem• 5 years ago

The biggest issue is that stories posted by people with high karma hit the front page even if they're off topic/dupes/garbage, while good new submissions often fail to get much promotion.

manx2• 5 years ago

Is that actually true?

If yes, instead of boosting all submissions, the user decide the boost, with quadratic karma cost, or similar.

nomilk• 5 years ago

This is great. I was suggesting a similar feature for search HN literally only a couple of weeks ago:

https://github.com/algolia/hn-search/issues/207

Looking forward to using!

mmiliauskas• 5 years ago

But can it predict the next financial crisis?

pjc50• 5 years ago

HN predicted 5,387 of the last two financial crises successfully.

onli• 5 years ago

Mapping the current points would be more interesting to me in practice. Right now, when you install the extension and read the frontpage you just see the points doubled, which is pretty useless. But if the extension would pick a fixed date - say 1.1.2020 - and then recomputes the data of older and newer submission with their exact inflation ratio relative to that date, then you would have a stable comparison point also useful for new articles.

furstenheim• 5 years ago

I've always thought something similar should apply to stackoverflow. Points from newer questions should be worth increasingly more, as more and more questions (specially the easy ones) are already answered

manx2• 5 years ago

Interestingly, the upvotes are quite inconsistent across multiple resubmissions of the same URL. Instead of adjusting against inflation, I think the HN formula itself should be adjusted: HNFormula(age,upvotes)/views. I'm working on a blog post about that, here is my current draft: https://felix.unote.io/hacker-news-scores

Show HN: Inflation-adjusted Hacker News

Related Stories