Tell HN: Don't ask for upvotes

[sama]

A lot of people ask us why their submission gets kicked off the front page. The answer is almost always that it trips our voting ring detector.

Please occasionally read /newest so interesting stories that don't solicit upvotes can still get on the front page.

[simonsarris]

Sam I think you have great power to give /newest more views if you'd consider minor redesigns.

Perhaps you could dither stories so the front-page list goes top/new/top/new/top/new, etc, but that is potentially very messy.

Or perhaps instead, the front-page can show 30 top stories, then a line break, then 30 new stories, all on the front page. Long-scrolling pages are in fashion, after all.

These aren't well-developed ideas, but the point remains that the inertia of being one click away means that a huge percentage of people will never even look at /newest, never-mind up-vote interesting stories.

Please consider changes to modify the median behavior. It's within your power and would do us all good. It's worth an experiment, isn't it?

[huhtenberg]

Just splice a random /newest entry into a front page on every load.

This gets extra eyes on them and helps float good stories up. Optionally mark them green, just like HN currently marks noob users.

[privong]

Another option is a 2-column layout (for non-mobile devices). The normal front page on the left, with the blast of new articles in the right column. Perhaps for mobile devices the new articles could go under the top stories, with some sort of obvious demarcation between them.

[sama]

good ideas here and in the responses. we will consider.

[chrisdevereux]

I like the idea of adding a sample of new stories to the front. It would guarantee a number of extremely low quality submissions on the front though, which might affect the overall percieved quality of HN.

It also might be a problem if simply submitting to HN guarantees a decent number of pageviews.

[tlrobinson]

I like the idea of simply appending a subset of the /new stories to the end of the front page. Make it clear they're brand new (should also be obvious from the low vote count).

[schrodinger]

I really like this idea. Just occasionally throw in some random "new" pages to users and see what gets read / upvoted. I'm sure that most users don't check the "new" page.

[lisper]

Some observations:

1. Almost no one reads the new page. For proof, go to the new page and count up the total number of votes that all the stories on the new page have received in the last hour.

2. There is a VERY short time window (typically about an hour nowadays) during which a story either gets enough votes to end up on the home page (half a dozen or so seems to be enough) or drops off the first new page, after which it will never be seen again.

3. Once a story arrives on the home page, there is a very good chance that it will stay there for a while (longer than an hour anyway).

4. As a result, there is an EXTREMELY strong incentive when one has posted a story to send out an email to one's friends saying, "I just posted this to HN, please take a look and upvote it if you think it deserves it." [1]

5. A story that gets a lot of votes as a result isn't necessarily a bad story. So a voting ring detector is likely to generate false positives.

6. Simply asking people to go read the new page isn't going to fix the problem. Figuring out what to do instead isn't easy.

---

[1] Last year I did an experiment where I wrote a series of six blog posts all related to a single topic. I posted all six to HN. Three of them I sent out an email announcement, and three of them I didn't. The first three all ended up on the home page, ultimately garnering many tens of votes. The other three never got a single vote.

[jsnell]

Low vote counts on the new page don't prove nobody goes there. An alternative explanation is that 90% of the new page is crap that really doesn't deserve a single vote. (Just did this experiment, and only found one of the 30 submissions interesting enough to even click on).

It is true that you need to get something like 5-6 votes in the first hour to get to the front page. But that's not a very large amount, and it's definitely possible to get across that threshold without soliciting for votes, so clearly there are a decent amount of people reading /newest. (About a third of my submissions made the front page, and it's kind of obvious in retrospect why many of the others failed. I never asked anyone to vote for any of those submissions).

[yesimahuman]

I'd be curious to hear more about how YC companies handle voting so they get their content on the front page. From what I've heard, they get a lot of help to promote content.

Totally makes sense seeing as HN is built around YC, but in that light makes it a bit disingenuous to ask the community to not solicit help from their group to promote content.

[tptacek]

I don't know what's "fair" or "unfair" in this context, but suggesting that Sam's comment is "disingenuous" is unproductive, because the voting ring detector isn't going anywhere. It's one of those things where most of us only see the occasional annoying side effect, but we don't notice the avalanche of crap that it's keeping off the front page.

You can probably safely assume that the decision to keep aggressively penalizing voting rings (and things that to the detector look like rings) is made and isn't changing.

With that in mind, read Sam's comment as "here's how not to get stung by the detector", not as an appeal to change your behavior for his sake. He's trying to help.

[sama]

YC companies complain the most often because they trigger the voting ring when they ask their batchmates to votes them up :)

[zt]

Other than "jobs" we get no help promoting content.

I actually think being a YC company can hurt you because all your YC buddies vote for you and then it appears that your cohort is a voting ring. I actually have to tell people NOT to vote for my posts sometimes, lest I trigger the HN Gods' wrath.

[dang]

> From what I've heard, they get a lot of help to promote content.

They don't from us (the people running HN). The only help HN gives YC cos are (a) being able to post job ads and (b) YC founders' names appear in orange to other YC founders.

YC startups routinely get hit hard by HN's ring detector.

[pbhjpbhj]

Perhaps promoted content should be labelled as such, eg an orange 1px border around the listing. If adding a new story in to the mix that could have a green 1px border to highlight it was out of normal flow too.

[kenrikm]

Just everyone in the office independently voting on something can trigger it, which is unfortunate since it was not even planned.

[louis-paul]

I think at this point, they have a network large enough to kick off the newest page to the front page pretty quickly.

[Mz]

On a full sized monitor, most titles go only about halfway across the screen. I think with a little tweaking, you could have two columns on the front page: One with the current front page code, one with the newest stories. You might have to have titles wrap for longer titles or make some other tweaks to make this work. But that might cut back on the need to ask folks to look at new stuff.

Generally speaking, if you can write the code, you can make the rules people have to follow (though it does require some deep thinking, data, etc at times). If you want to compel x behavior, then compel it. If you want to forbid y behavior, then deny the ability in some way where possible. Writing the code makes you "god" for your little world: You decide the local "physics". Deciding the local physics is going to generally get you better results than asking people to please do x or please don't do y.

[webhat]

I did something similar for an RSS reader I was building based on bayesian filtering, a col-4 column on the right with the "live" feed and a main col-8 column with the filtered feed, meaning all the matches. It looked pretty good in landscape mode.

[gus_massa]

I usually read the newest page, but usually it has a lot of 1-point stories, and many of them are almost-spam or almost-dupe.

Another user proposed a newest page with a >2-points filter (someone else think that it’s interesting). See for example http://hnapp.com/filter/d3a308f2ac9a071c0bf174e0c1a8fd22

[ColinWright]

But that would mean that stories with only one point never get seen, and hence never get up-voted, and hence nothing gets to the 2-point page. It seems necessary that people should read "newest" and up-vote things they think are of value.

[james33]

How does anything get on that list then if nobody ever sees the 1-point submission (without asking for votes).

[weinzierl]

When I read the newest page I usually ignore the points. Most of the stories are 1-point or 2-point anyway. What I scan for is the comments. If the story is on newest and already has a few comments it's often worth reading in my experience.

[oskarth]

> Please occasionally read /newest so interesting stories that don't solicit upvotes can still get on the front page.

In theory this is good, but in practice it's hard. Here's a suggestion for how to make it easier in practice: redirect 1/3 of the visits from the start page to /newest. As a regular and medium-karma user, I would know why this happens, and I personally wouldn't mind it.

[sharemywin]

I prefer /newest. I'd rather see stuff unfiltered. Plus you can comment on stuff without 150 other comments. so it feels like your comments are heard.

[lucb1e]

I think they're read more than one realizes even if it doesn't gather upvotes

[compare]

Is there actually any dataset of upon which the accuracy of this detector is measured and improved? Or is it just some hard-coded rules?

For example, does it really differentiate between people who ask for votes, and people who just vote up the content of someone they recognize?

[aikah]

What does count as a voting ring?

If I post a link to HN on twitter and people upvote the submission coming from twitter,does it count as a voting ring?

[tptacek]

You're never going to get a totally straight answer, because things like voting ring detectors rely in practice on obscurity; if you knew the exact details, it would be much easier to game the system.

[rickyc091]

It's probably better to up vote from the feed. When you click on a link from the feed, you're generally going directly to the article. You hit the browser back button to get back to hackernews; chances are if the article seemed interesting, you would then hit the up vote button. When's the last time you hit the up vote button on a comment page?

[mathrawka]

Always the same people giving upvotes to a certain user.

[encoderer]

Of course, the alternative is just to submit something to HN and..... hope for the best? Sure, THAT sounds reasonable to a bunch of type-a founders.

I think it's reasonable that somebody promotes their HN post. The takeaway is just to be careful about how you share and how widely and with whom.

I made the front page last week (for a few hours) for a side project that I launched with a friend. Several up-votes shortly after we posted it got us to the bottom of the front page, and organic traffic took it from there.

[programminggeek]

I remember the early days of digg and people tend to see social news sites as a mechanism to get a wave of traffic. What they don't always think about is providing value to the social sites.

HN is probably better than most about not letting much crap on the home page, but these sites are almost universally terrible at surfacing new content so that it has a good opportunity to get up voted.

The selfish human response is to get your friends to vote for you (just like you get your friends to vote for you in real life actually). It's just that there are so few actual people looking at the submitted stories that it wouldn't take much to astroturf your way on to the front page without a voting ring filter.

The best solution is very likely to have a better interface for surfacing submitted stories to a wider audience. Or, incentivize people to actually look at the new section. Maybe extra points for voting or commenting before it hits front page.

I don't think it would take much to solve this problem, but this isn't a policy or filter problem, it's a human problem and your system needs to work with the user psychology that naturally exists on HN (and similar sites).

Reddit seemed to solve this by making subreddits a bigger part of the site, but I don't know how HN would incorporate such an idea without losing the site identity.

[TeMPOraL]

This post just motivated me to go through /newest and upvote few things. Thank you.

Everyone, let's take a minute now and go through the new submissions and move some fresh things onto the front page. Let this also become a habit, until mods tweak HN interface to reduce the effort it requires.

[PhasmaFelis]

If legit submissions are getting booted by your voting-ring detector so often that you need to ask readers to take action to prevent it, then the detector is broken. It's not the readers' responsibility to change our habits to make up for your broken mechanisms.

[Kiro]

I wonder how many false positives there are.

[schodge]

If the OP is still reading, why didn't this at least get onto the Ask HN page? I wasn't asking for upvotes, but it seemed to fall like a stone without ever having a chance to get views... ?

https://news.ycombinator.com/item?id=7948001

[arikrak]

Currently, it seems you need either a lot of good luck or upvoting friends to make it to front page. Why not use some categorization so every link makes it it's category's page and has chance of reaching front page?

[lbr]

Sam, Does saying "discuss on hn: link" at the bottom of a blog post violate your voting ring detector or any other terms of service? I couldn't find anything about that.

[mhartl]

What do you think about soliciting upvotes by sending people to /newest? Among other things, as a side-effect they might upvote other stories as well.