Hacker News Reader

Comment by jacquesm | Hacker News Reader

Shall we put it to a vote ?

Even if people said they'd prefer I focused on adding features like search rather than the quality of submissions and comments, which is unlikely, I still wouldn't.

I find it hard to believe that a programmer of your caliber would make it seem as though an off-site submission is such a big deal that it would take away significant time from working on the quality of the submissions and comments.

I'm not 1% the programmer that you are and it wouldn't take me more than 15 minutes to put a search box (after all, it's just a piece of static text) in the HN footer.

See, all those 'quality submissions and comments' are pretty much useless if you can't find them, what's the use of having this amazing body of content if there is no easy way to peruse it?

The content goes by so quick now that if you look the other way for three days that you'll have missed tons of good stuff, an easy way to search would make all of HN accessible, not just what is current right now.

What you don't get is that attention is a zero sum game. There are an infinite number of minor things that would be pretty good to do that would "just take a minute." The only way to do good work is to ruthlessly blow them off and to focus only on the few things that matter most.

Ironically, this thread, though a complete bike-shed waste of time in itself, is not entirely bad, because it's also causing me to spend time thinking about the kind of problem I should be thinking about. One of the big questions I've been mulling over is what to do about points on comments. I'm increasingly inclined not to display them, because an upvote or downvote is equivalent to a very inarticulate comment. And this subthread is an argument for doing so, because the only reason I felt obliged to respond to you is the number of points your comment got. If points weren't displayed I wouldn't have felt I had to respond.

Basically, points exacerbate the "someone is wrong on the Internet" trap.

Found using searchyc.com:

http://news.ycombinator.com/item?id=740983

(google doesn't work well when you search for words in the UI of HN)

Apologies for the frankness. I keep trying diplomacy but I suck at it.

A less informed person might think that since you got your clock cleaned by popular vote, you seem much more inclined to discount popular vote than you did a month ago.

I have lost much of my admiration for HN. But I have an awesome appreciation of what you think is important to spend your time on. I have to admit I am mostly wrong and I am still trying to learn as much as I can from you, but why the hell do you think here that Jacques is giving you a bad requirement for your site? You sound pompous and disconnected.

EDIT: And am I to understand you monitor the voting records on HN? That sounded crazy, but I just heard it via email. The internet is crazy. Can't be true.

Comment and community quality are such hard problems to solve, and are orders of magnitude more important for a site like this than features like search. Sure, it might be easy to add search. Or rankings. Or a million other things a small minority would appreciate. But that's not how you make a good site that is optimal for the larger population. An option should be very useful for a large percentage of users to make it worth adding to the complexity of something. Users say they want every feature under the sun, but after a lot of that, when they see something well designed like the iPad, and they're generally struck by how nice it is to be free of that grabbag design philosophy.

Speaking anecdotally, I've tried to dig up a past item maybe 5 times in the past year via search, and none of them were very important. Compare that to the maybe 300 times I've been back to the site, solely for the content quality. I could do without HN searchability altogether very easily, I very rarely reread things. I would venture that this is a common usage pattern of a news site with ephemeral content.

You sound like an entitled and disconnected user, asking why he's not adding something so simple that you personally want, rather than thinking about what would make this site better for everyone. Search is not going to make the site better for most, and people that respond to a vote about the subject are not going to be the apathetic majority.

The argument PG brought was that it would take time away from the other work. But as it is it incrementally takes time away from all of us, and all the time.

The spending of that one or two minutes at integrating searchyc.com or google.com would come back 1000 fold over the life time of the site, and probably much more than 1000 fold, and if he's that busy he maybe should pass the torch to people that can fully concentrate on making HN, the best site of its kind right now even better.

I personally find it very hard to believe that the trade-off is as stark as is portrayed here, working on a site like this is not in 'absolutes' like that, you simply do stuff because it has to be done. If a YC funded start-up would display that attitude to their users they wouldn't stand much a chance, in fact it is exactly opposite to what Paul says in other places to how one should deal with users.

It seems more like a stubborn 'I'll do this my way' kind of thing, Paul has kept the door solidly locked to others contributing to HN, but then goes and makes a big case about how he has only so much time to spend. It's hard to be arguing both of those at the same time.

Anecdotally, I've used search 100's of times to find a comment or an article that I'm pretty sure I read on HN but at the time didn't have an application for so I didn't bookmark it. This happens all too often in the tech sector, after all you can't know in advance which technology will land on your plate 3 months from now.

"Paul has kept the door solidly locked to others contributing to HN"

Does he? If you send a bug fix or feature patch to the HN arc codebase, he tosses it without looking at it? Or does he look at it and then take a decision whether to include it or not? I would imagine the latter (please correct me if I am wrong).

That he refuses to consider some suggestions, no matter how much sense it may make to the person making the suggestion (or even bystanders) is the exact same behaviour that every open source project's BDFL exhibits. I think you exaggerate with the "door solidly locked to other s contributing". If you send in a bug fix patch for example, I am sure he'd incorporate it asap.

Now the complaint reduces to "but PG refused to consider this feature though I (and many others) think it is a must have"

The traditional answer to "but the BDFL refuses to incorporate my suggestions which were liked by all the users I spoke to" is "then fork the code, and/or build something better".

My view, fwiw, is that we (users of HN) have the right to request features and present a logical case and PG (as the chief programmer/owner/BDFL etc of HN) can accept or refuse those requests for any reason whatsoever. If he explains the rationale that is a bonus, but he doesn't really need to. It is his project.

If he refuses to incorporate our fixes/suggestions, we (the hacker users of HN) can either go along with his decision xor fork the codebase (or start a new project from scratch using our preferred tools) and build something better (and I know a couple of HNers who are trying exactly that).

http://news.ycombinator.com/item?id=830736

'nough said. I've been asked 'from on high' to stop this thread and others like it so I will.

Keeping a vision intact often does require absolutism. It's helpful to experiment, but in the end, I think a maintainer has to stick to his guns to keep anything from devolving into crap. Guidance from users is often stunningly wrong when taken from a higher vantage point. I hate to keep running back to Jobs, but if he listened to popular opinion and let them guide what he wanted, his products would end up reverting to the mean. Instead he keeps his door "solidly locked".

To implement search well is not a trivial task at all, and distracts from the much, much more important task of making the community self regulate towards quality. If all you want is a link to site:news.ycombinator.com, then a bookmark should be fine for you and others who do it frequently, and typing it in should be fine for those of us who search infrequently. No additional site complexity needed. If people ask how to search it, then they'll learn a neat and generally useful trick when someone tells them about site:xyz on Google.

Edit: And in response to "he should bring on more people if he's so time-limited", bringing on more people frequently causes more of a time drain than it solves.

I read HN multiple times a day and have for a while now. I do not care about search at all. This site doesn't really have any features, and I don't care - I come here for the community. Just another data point fwiw.

Does it really take more time to type "site:news.ycombinator.com my query" into the OmniBar than it does to find a text box on the page and type "my query" into it?

I've used search 100s of times as well to find a comment or article that I previously read, but I just Google it. I don't see how having a dedicated search box would be any faster. (And if he used Lucene or a custom algorithm instead of offloading things to Google, it'd probably be significantly worse.)

"And am I to understand you monitor the voting records on HN?"

As revealed here: http://news.ycombinator.com/item?id=1278680

From that post, you can't conclude that pg "monitors the voting records" as a general practice, only that he's willing to look at them when someone specifically complains about the ranking algorithm. Read the parent comment to that one -- the guy complained about what type of stories were on the front page; pg made the point that the complainer wasn't doing his part to get other stories up there.

I don't think that Mahmuds' point was about the ranking algorithm, it was about the submissions themselves. As I pointed out in that other thread the fact that he didn't vote for those stories was his whole point, to quote his voting record back in public to chew him out is imo really below the belt and was - at least to me - totally unexpected.

I think the points helped you gain a more accurate idea of whether responding was worthwhile. The more people support an idea, the more response it calls for.

Furthermore, an upvote or downvote is equivalent to a very concise comment.

The base case of concise and inarticulate are identical.

The second most concise form of comment would be to say simply "yes" or "no." Do you feel such comments add to a discussion on account of their conciseness?

Comments like yes, no, agree, disagree, good post, and lame are concise, but also boring wastes of time which add little to a discussion. The aggregate measure of those sentiments (as expressed in a point total) is both more concise and of some value, as it serves to encourage and highlight good posts and discourage thoughtless or pointless comments. It also provides a relatively unobtrusive outlet for those who wish to voice appreciation of thoughtfulness; this is preferable to the sea of "agree"s and "disagree"s that might spring up in the absence of a point system.

I feel such comments when presented like news.yc votes or FriendFeed likes add more value because of their conciseness.

  (define (value conversation-element)
    (- (benefit conversation-element)
       (cost conversation-element)))

  (define (cost conversation-element)
    (- (conciseness conversation-element)))

I'll leave the definition of (benefit conversation-element) as an exercise for the reader.

I'd be careful about messing with points on comments (and with karma more generally).

You might enjoy imagining a forum where smart people anonymously engage in civilized discussion without need for the kind of lowbrow pleasure-center-stimulation that the current karma system provides, but I doubt such a thing could exist; if it did, it would probably be much less exciting than HN. Smart people who enjoy good discussion are still people, and most of what people do is seek out easy rewards.

You've built something almost impossible here: a place where it's very rewarding to say interesting things and very unrewarding, generally, to say uninteresting things. Don't look this gift horse too directly in the mouth. If you kill the point display on comments I think you'll kill a lot of the content on this site.

Not to mention that, though it's easy to make fun of people who get too sucked in to internet arguments, there's frankly nothing wrong with people going out of their way to correct other people and that kind of behavior is pretty fundamental to crowd conversations like HN. It's probably annoying that jacquesm's popular approval forced you to respond to him, but I (and others) enjoyed reading your comment; I think you'd be hard-pressed to show that you writing it was a net loss activity.

LambdaTheUltimate? No points there, and it's resisted the influx of trolls better than most.

Basically, points exacerbate the "someone is wrong on the Internet" trap.

This absolutely squares with my experience. I wouldn't want to lose visibility into my own comment scores, but I could do without seeing everyone else's.

> There are an infinite number of minor things that would be pretty good to do that would "just take a minute."

The search link is at least as relevant as the bookmarklet link and certainly more relevant than the webmynd and mixpanel links with images.

> Basically, points exacerbate the "someone is wrong on the Internet" trap.

While i am not sure about that, I suggest that voting for comments should be delayed like the temporarily hidden reply link, that may reduce some reflexive voting.

I like the idea of getting rid of points. I dislike the idea of losing the ability to quickly scan for comments the community thinks are good. Because of the clever way new comments are briefly at/near the top of the page, something should stand in for "this is probably worth reading".

Make something people want, right?

If the majority of folks here want search, and given that it wouldn't take much time to add a link to SearchYC, why not do it?

The reason I wouldn't add a link to SearchYC is the same that I don't add a link to one of the HN iPhone apps: I don't know the authors or the app well enough to anoint it as official. I don't mind if people want to build stuff on top of HN, but I have a much higher threshold for making something part of HN.

Searchyc is at least as reliable if not more reliable than HN, I think you'd be in pretty good hands there.

In all the time that I've been using it I think I've seen it down only twice, HN over the same period numerous times.

HN handles considerably more traffic than SearchYC.

Well, that's a difficult one: Either PG is right and nobody needs search, then we'll find it doesn't matter, or I'm right and we do need search and then the searchyc guys will need to beef up their system.

But then at least we do know for sure we need that search.

As it stands, SearchYC doesn't get enough daily uniques to justify beefing up the system. Frankly, SearchYC was an experiment in how quickly I could learn a new technology (RoR) and build a system. I was completely expecting someone to come along a few months later to make SearchYC obsolete. It's been 2.5 years, and I don't think that's happened yet.

If you got a lot more traffic because YC would anoint you the 'official' YC search engine would you be able to handle it?

Would you commit to beefing up the system to whatever it took?

Well, no. SearchYC has no revenue. I maintain SearchYC and host it out of a box in my closet as a service to the community. It doesn't make any sense for me to commit to do "whatever it takes".

That being said, with a few hundred dollars worth of hardware and some software changes, SearchYC should be able to gracefully handle a couple orders of magnitude more traffic.

Consider yourself funded if it should come to that.

Bravo: this is the first time I've seen you publicly acknowledge that they exist. Baby steps!

You know them well enough to invite them to a private party at your office. Do you know the CO2stats people better, given that you're including a cross-site web-bug javascript from them that frequently blocks the site's page rendering?

It's a few hundred bytes of <form>, you could remove it later if they flamed out with the same ridiculous ease with which you could add it.

Reply on news.ycombinator.com