zlacker

[parent] [thread] 27 comments
1. dang+(OP)[view] [source] 2020-04-03 08:17:42
Ok, since this is getting attention, I'll explain.

Launch HN posts for YC startups are one of three formal things that HN does for YC (https://hn.algolia.com/?dateRange=all&page=0&prefix=false&qu...). The others are job ads for YC startups, and orange usernames for YC alumni—but only when displayed to other YC alumni, which always generates "why is my username not orange" emails. But I digress.

Launch HNs are like job ads in that they get an initial front-page placement, usually somewhere between #8 and #10. (I think job ads start a little higher). Then they fall down the page. Unlike job ads, though, launch posts can be upvoted and commented on. Once they've gotten their initial placement they function like regular stories. Occasionally the community finds one particularly interesting and it gets upvoted higher. This recent one spent quite a while at #1: https://news.ycombinator.com/item?id=22616857 (http://hnrankings.info/22616857/).

We started doing the Launch HNs three years ago. I was worried that the community would hate them because we were taking additional front page space for YC. (Our intention was to make it so that a launch post and a job ad wouldn't appear at the same time, but I never ended up writing that code, so sometimes they do.) But that hasn't ever come up. I think it's because launch threads are intrinsically more interesting than job ads (see https://news.ycombinator.com/item?id=22767319 downthread for more on that).

All this time, the initial front-page placement of launch posts, unlike job ads, has been done manually. That is, founders have had to email us and we've manually jigged the post onto the front page. The problem with that is that you have to be awake to do it. I don't want to be awake to do it, especially because startup founders tend to be businessy, bustling types who are all bright-and-early, while my schedule drifts ever deeper into the darkness as the few moorings I had remaining to the rest of society dissolve in this time of social distancing and self-isolation, which were basically my bread and butter to begin with (https://www.youtube.com/watch?v=yUlBWNDW72E).

Until now, I've told founders to post at Pacific 10am and email us, because that's roughly when I get going in the morning. Tomorrow, though, there are two. One is a fintech startup in Latin America who want to post at 9am. And the other is Peter Roberts, who's going to do another immigration AMA (https://news.ycombinator.com/submitted?id=proberts). Peter is on the east coast, and wants to get going at 8am Pacific, which is late for him and (ungodly) early for me.

So tonight I got frustrated enough to write some code to deal with it. Frustration reaching a tipping point and boiling over is my gateway into the code these days. My goal is for startups to be able to post their launches, and occasionally for pre-scheduled submissions like Peter's (which are rare), to end up on the front page independently of whenever I went to sleep the night before.

This code turned out to be a lot more complicated than I anticipated. The patch ended up adding a hundred lines of Arc. A hundred lines of Arc! Do you have any idea how many lines of Arc that is? I just looked through the history and the last commit that added that many lines of code was over two years ago when we got Arc to compile to JS. Obviously this change needs to be thoroughly tested, so after testing it on my laptop I deployed it to production and decided to do a couple of sanity checks live. One was to post a test Launch HN using my old account gruseom, which is the founder account for Skysheet, the spreadsheet startup that Scott and I had 10 years ago (and which I still think about every day, but I should avoid digressing again). The code I wrote has some logic in it for cofounder accounts. One thing it's supposed to do is email all the cofounders when a Launch HN post has made it to HN's front page, so they will be ready to engage with commenters. Anyhow, this test post was the OP. The good news: it got placed on the front page in the way that I intended. The bad news: none of the emails were received. God fucking shit fucking goddamnit I knew those emails wouldn't work...I mean, dang.

(Edit: actually that's not what happened. https://news.ycombinator.com/item?id=22772667)

My intention was for this thread to remain obscure, then get placed briefly on the front page by the new code, at which point I'd get the emails and immediately delete it. I didn't think it was very likely to get noticed in the middle of the night here, especially when posted by what ought by now to be an obscure account, but oh well.

I think that covers everything. Hopefully you all will see those two auto-placed posts on the front page tomorrow morning because I do not intend to be awake 5 hours from now. (Edit: it worked!)

replies(6): >>fyfy18+N >>soneil+2a >>Aissen+kj >>jboric+yF >>krapp+hV >>thunde+Jj1
2. fyfy18+N[view] [source] 2020-04-03 08:27:28
>>dang+(OP)
One thing I always wanted to ask if why do job ads not allow comments? I'm guessing its to prevent negative comments/trolling (we get plenty of 'this will never work' on Launch/Show posts), but at the same time I feel like comments on job ads could be a good way for the founders and existing employees (tell us why you as a meager developer are excited to work there!) to interact with the HN community.
replies(2): >>_-____+l1 >>dang+R1
◧◩
3. _-____+l1[view] [source] [discussion] 2020-04-03 08:32:46
>>fyfy18+N
Job ads in the "who is hiring" threads generally only attract comments of one of a few types:

"I applied to this place and they never got back to me"

"Thanks, I applied"

"Please add salary/remoteness/interview process to your job ad"

"You have a typo in your text/email address/website URL"

replies(2): >>h0h0h0+U1 >>stevek+a7
◧◩
4. dang+R1[view] [source] [discussion] 2020-04-03 08:39:04
>>fyfy18+N
What _-___________-_ said. The short version: job ads are boring. The long version:

If job ads were threads, each thread would be a generic referendum on the company. Worse, it would be the same referendum over and over.

Job ads are boring, so there wouldn't be anything to discuss other than how one feels about the company, and that's boring except to people who have strong feelings on the topic, and strong feelings on the internet tend to be negative, so the threads would fill up with negative generic comments.

I believe that the hivemind resents boring things, such as submissions where the only new information is "X is hiring", so it gets cranky and fills the vacuum with indignation, basically as the only way to amuse itself in the absence of anything interesting to discuss. It doesn't want to, it just doesn't know any better way to have fun in a vacuum. In practice, what this would look like in a job thread is "I applied in 2015 and never heard back", plus—if there has ever been a negative story X about the company—every variation of X, X, X, repeated increasingly snarkily.

Actually, it's worse. Building a business is a long hard slog. One needs to hire more often than one has scintillating new information for the community to have fun discussing. Therefore, each successive job ad would be even more boring than the previous one, leading to monotonically increasing resentment. Repetition is the enemy of curiosity.

Launch posts don't suffer from this dynamic because by definition, the startup is new, so there's something new to discuss.

◧◩◪
5. h0h0h0+U1[view] [source] [discussion] 2020-04-03 08:39:21
>>_-____+l1
The latter two comments seem pretty useful feedback though
replies(2): >>dang+R2 >>bryanr+ab
◧◩◪◨
6. dang+R2[view] [source] [discussion] 2020-04-03 08:50:53
>>h0h0h0+U1
But not interesting.
replies(1): >>laumar+W4
◧◩◪◨⬒
7. laumar+W4[view] [source] [discussion] 2020-04-03 09:18:02
>>dang+R2
[edit: before you read on bare please in mind I'm not advocating comments be enabled (like many others have assumed).]

I think it's a little redundant making a distinction between "useful" and "interesting" because either way the comment has value. However I do agree with the points you made in your other comment (https://news.ycombinator.com/item?id=22767319) and that the examples given of useful/interesting comments by the GP doesn't offer a high enough value to justify the inevitable negative and other low value comments. Which I think is the real crux of the matter. Much like why political discussions are generally banned on here, the signal to noise ratio just isn't worth the few valuable comments a submission might receive.

replies(1): >>411111+L9
◧◩◪
8. stevek+a7[view] [source] [discussion] 2020-04-03 09:48:27
>>_-____+l1
And posting in any of the "Who's looking for work" threads mostly just results in lots of people sending you emails:

"We saw you post, would you like to join our platform ..?"

Lots of upwork knockoffs; I've started sending them GDPR Subject Access Requests to see what the companies have scraped and held about me. Next step is obviously data-deletion requests. Fun.

replies(1): >>dang+Z7
◧◩◪◨
9. dang+Z7[view] [source] [discussion] 2020-04-03 10:00:09
>>stevek+a7
They're not supposed to do that, as the rules at the top of the thread make clear. Perhaps we should build some kind of 'flagging' feature so people can report bad actors. Then we could add a warning that bothering HN users illegitimately will land them in the bad dog box on HN.
◧◩◪◨⬒⬓
10. 411111+L9[view] [source] [discussion] 2020-04-03 10:20:02
>>laumar+W4
Nobody is stopping anyone from providing that feedback through mail, so still no need for comments.
replies(1): >>laumar+Eb
11. soneil+2a[view] [source] 2020-04-03 10:25:21
>>dang+(OP)
Possibly the most interesting test post I've seen in a long time, thank you. I suspect the "current circumstances" worked against you here - you have no idea how bored Europe was by 10am.
◧◩◪◨
12. bryanr+ab[view] [source] [discussion] 2020-04-03 10:42:15
>>h0h0h0+U1
number 1 is pretty useful feedback for me, if they don't respond to people who apply I don't intend to take time to make a good cover letter and adapt my CV to highlight how I would be a good fit for the job.
replies(1): >>laumar+oc
◧◩◪◨⬒⬓⬔
13. laumar+Eb[view] [source] [discussion] 2020-04-03 10:50:42
>>411111+L9
I don't understand why you're downvoting me when you're ostensibly making the same point I was. Did you actually read my comment to the end or just made the assumption I'm wrong because I replied to dang? Because I was actually agreeing with him for the most part.
replies(2): >>_-____+sc >>411111+Fj
◧◩◪◨⬒
14. laumar+oc[view] [source] [discussion] 2020-04-03 11:00:27
>>bryanr+ab
As a hiring manager I can assure you that is not always practical. If someone makes it to an interview, even if it's just a first stage telephone interview, then I'll readily give feedback. If some sends a CV and then asks for feedback then I will also happily provide feedback if time permits). But recruitment is a heavily time consuming process already and some positions can receive dozens or more CVs so I don't have the time to reply to every single candidate and explain to them "Thank you for applying but unfortunately we've had better CVs through." Likewise I've never expected that when applying for other jobs either.
replies(1): >>bryanr+Kd
◧◩◪◨⬒⬓⬔⧯
15. _-____+sc[view] [source] [discussion] 2020-04-03 11:02:02
>>laumar+Eb
The distinction between "has value for the submitter of the job ad" and "has value for the entire community" is an important one. Comments that only meet the former test can just be sent directly to the submitter by email or whatever.
replies(1): >>laumar+dd
◧◩◪◨⬒⬓⬔⧯▣
16. laumar+dd[view] [source] [discussion] 2020-04-03 11:11:47
>>_-____+sc
Preface: before you read on bare please in mind I'm not advocating comments be enabled (like many others have assumed)

> The distinction between "has value for the submitter of the job ad" and "has value for the entire community" is an important one.

I disagree. For example if there is a typo in a URL could take a while for the poster to update it and they might not even be able to edit their submission if it's not identified in time. So if a community member says "link to xyz.com should by xyz.org" that would help other community members who wouldn't know what the correct URL was meant to be. That kind of comment has real value to everyone and not just the submitter.

> Comments that only meet the former test can just be sent directly to the submitter by email or whatever.

Lets be clear (because I've already stated this twice and it still seems to be overlooked) I'm not advocating that comments should be enabled. I agree with the point that those kind of comments are better off sent out-of-band because they don't offer enough value to justify the inevitable low value comments that would also follow. The only point I was making was that the distinction between useful and interesting shouldn't be overstated because they are typically (though obviously not always) linked.

◧◩◪◨⬒⬓
17. bryanr+Kd[view] [source] [discussion] 2020-04-03 11:20:16
>>laumar+oc
ok, but I would think like this:

looking for a job is a heavily time consuming process, and going to a new job is a risky process. I'm too well-situated to take that time or risk on any thing that seems off. I would have to be desperate to change that calculation.

If a company does not reply I assume there are potential reasons:

1. company is disorganized, just as a company would penalize me for seeming disorganized I will certainly do the same with a potential employer. The point of a company is in some ways to be more organized than individual humans; it is, after all, an organization. If it can't or won't be organized I won't have anything to do with them.

2. Company is rude. Treating someone badly when you have no power over them is a warning sign never to let the company have power over you.

3. Company does not have good tools setup to automate responses to people whose applications it has decided not to go further with - which is a subset of company is disorganized.

So I guess there is a mismatch between our goals and needs in the requirement process.

replies(1): >>dsr_+Yl
18. Aissen+kj[view] [source] 2020-04-03 12:16:23
>>dang+(OP)
"middle of the night here" is morning for most of Europe. Quite late in the morning, when people had time to work and maybe do an HN break. I've noticed that stories posted in the morning (for Europe) can get as much traction as other in the evening (american morning); especially if they touch to subjects we are quite sensitive to.

Did you do any study on HN stats and where its audience is coming from ?

replies(1): >>dang+5r1
◧◩◪◨⬒⬓⬔⧯
19. 411111+Fj[view] [source] [discussion] 2020-04-03 12:20:22
>>laumar+Eb
> I don't understand why you're downvoting me when you're ostensibly making the same point I was.

I didn't. I couldn't even if I wanted to, as I always create a new account once downvotes are unlocked as I don't like the temptation of destroying discussions for topics I don't agree with.

◧◩◪◨⬒⬓⬔
20. dsr_+Yl[view] [source] [discussion] 2020-04-03 12:39:41
>>bryanr+Kd
#3, especially: if your company has sane mail tools, it can autorespond to every single application with a variant on this:

Thanks for applying to $COMPANY.

We will get back to you by $(TODAY + 7) if we want to start a conversation.

Sincerely,

A. Robot

replies(1): >>within+Gr
◧◩◪◨⬒⬓⬔⧯
21. within+Gr[view] [source] [discussion] 2020-04-03 13:28:27
>>dsr_+Yl
I generally give companies a week or two to reply. If they don’t ever reply, I never apply there again. Or I’ll apply and ghost them years later. Several times.
replies(1): >>throwa+qs
◧◩◪◨⬒⬓⬔⧯▣
22. throwa+qs[view] [source] [discussion] 2020-04-03 13:33:42
>>within+Gr
This thread is a great generic example of why job posts don't enable comments.
23. jboric+yF[view] [source] 2020-04-03 14:54:57
>>dang+(OP)
> Frustration reaching a tipping point and boiling over is my gateway into the code these days.

Couldn't have put it better myself. Thanks for the work you do!

24. krapp+hV[view] [source] 2020-04-03 16:14:04
>>dang+(OP)
>This code turned out to be a lot more complicated than I anticipated. The patch ended up adding a hundred lines of Arc. A hundred lines of Arc! Do you have any idea how many lines of Arc that is?

It's like a million!

replies(1): >>dang+Kq1
25. thunde+Jj1[view] [source] 2020-04-03 18:03:30
>>dang+(OP)
Dang,

Yours are the only comments I always read.

Thanks for all the stuff you do. Wouldn't have thought you coded too! That's good to know.

I've also had to do tests in production occasionally and hope nobody notices! Good that your code wasn't tied up to the number of upvotes, otherwise you would have a different problem on you hands!

replies(1): >>dang+RP1
◧◩
26. dang+Kq1[view] [source] [discussion] 2020-04-03 18:43:38
>>krapp+hV
Exactly! 10000x better than blub.
◧◩
27. dang+5r1[view] [source] [discussion] 2020-04-03 18:45:07
>>Aissen+kj
Oh indeed, it's eternal morning on HN. Or rather, eternal second breakfast / coffee break. I did not mean to imply otherwise. It's just easy to forget when in dev mode and it's dark outside.

> Did you do any study on HN stats and where its audience is coming from?

I did that 2 years ago: https://news.ycombinator.com/item?id=16633521

I keep meaning to rerun that analysis. I saved the code for it using the trick described here: https://news.ycombinator.com/item?id=22516834.

◧◩
28. dang+RP1[view] [source] [discussion] 2020-04-03 21:48:51
>>thunde+Jj1
I believe that it has been important to HN's evolution that the (full-time) moderators and programmers have always been the same people, and that we should preserve that invariant.
[go to top]