Hacker News new | past | comments | ask | show | jobs | submit login
“We’re winding down Starfighter” (twitter.com)
112 points by j_s on Sept 2, 2016 | hide | past | web | favorite | 32 comments

Who was the "twerp" tptacek referred to? That thread is confusing, that the submission is gone doesn't help.

I think he's being cute. Try replacing that string with "tweet".

Yes. That is his word for tweet. I'm glad one of us thinks it's cute. :-)

Thanks. Let's be patient to discuss this until there's something substantial to discuss: https://news.ycombinator.com/item?id=12410410.

Again: I have stuff to say here, both about what I learned about starting a recruiting business on this model and interacting with people's hiring models, and about what I'm doing next with the challenge-type stuff, and about what Erin and I are in general doing next. This is just a tweet. Is there that much to talk about? Give me, like, maybe a couple days?

I'll leak a bit of what I have to say here on this thread:

You would think that having anything you say about your company rocket to the top of HN would be an immense blessing, and maybe it is and I'm just too dumb to capitalize on it. But my day-to-day usage of HN, the way I talk about things here, and the way I comport myself are anything but strategic, and I found the attention we got on HN to be incredibly difficult.

Finally: I'm not done. I'm just done with this specific iteration.

> I found the attention we got on HN to be incredibly difficult.

Now that's how you stoke curiosity.

I though Stockfighter was a cool idea, and was incredibly excited to try it when it launched. Unfortunately the site was unreliable, winning felt random, and in the end I'm not sure that I would have had faith that it was a good litmus test for developer skill.

It was a great effort, but I feel like it never really caught on and that a community never really developed around it.

As a comparison, I found Elevator Saga [1] to be much more approachable and fun.

[1]: http://play.elevatorsaga.com

I had fun with the first few levels, but like you I noticed little difference with better or worse strategies - it all felt pretty random.

As I got into the harder stuff, I quit because it just felt like work for me and the last thing I want to do at the end of the day is more coding.

Did you attempt the second set?

The embedded device levels? Or were there levels after 6 for stockfighter?

I played around with stockfighter in December/January and in July-ish. While the service was more stable in July, the lack of documented endpoints for level selection or the back office was still a bit of a turn off.

There were a few things I found most frustrating:

- the trading bots concentrated most of their activity in a 5-10 second chunk every 60 seconds. This made it hard to "react" to trading behavior in a meaningful way throughout the "day" when my latency was in the 150ms+ range. I think this was supposed to be part of the challenge, but instead I found it to be an odd approach to getting people engaged.

- I spent a lot of time on level 2 trying to code a bot that could win every time. Maybe I just don't understand HFT (very likely) but I often felt powerless to impact the result and instead was a victim of the whims of the bots. I could beat the level, often handily if the bots traded my way, but I never reached a point where I felt my actions were meaningful. It felt less like an exercise in good engineering and more like a game of chance. And maybe that was meant to reflect the real world, but it certainly didn't feel like a game where I was actually increasing in skill as I progressed.

In reality, I don't think CTF style games are really for me. But I do love a good engineering challenge, and really loved the blog posts about how everything was put together. I wish I could have been doing that work instead of trying to outwit the system. Building a system that fulfilled an API might have made me feel more satisfied than the other way around.

Are there successful startups working on what you may broadly call "programming games (with a serious practical purpose)"? I know of CodeCombat [1]. The story of how they got into YC [2] is pretty interesting:

>With all the chaos trying to keep the server up and the bugs down, we slept little and prepared for the next day's Startup School even less. We had been tapped for on-stage Y Combinator office hours with Paul Graham and Sam Altman. We watched a video of previous on-stage YC office hours and concluded that "office hours" really meant "eight minutes of two of the smartest startup guys in the world demolishing your idea in front of 1700 entrepreneurs and a live video stream".

>Fortunately for us, they liked our startup and were much nicer than we expected. In fact, as we were walking off stage thinking, "Hey, that went well--maybe we'll get an interview!"--then Paul whispered something to Sam, who nodded, and they called us back.

>"Okay guys. Wait, wait, come back! Come back for a second. You didn't realize that, but that was your Y Combinator interview. You're in the next batch."

It's worth watching the video for the founders' entire interview.

Edit: Added the quote.

[1] https://codecombat.com/

[2] http://blog.codecombat.com/codecombat-in-y-combinator

There's codingame[0]. It teaches video game programming through "programming games". They're pretty fun, too.

[0]: https://www.codingame.com

Well that's too bad, it was a brave attempt at trying to fill a general recruitment need. I hope that the lessons learned can go into the next round of recruiting tools.

I've finished both microcorruption and jailbreak (at least, the publicly available bits). I really enjoy these single player CTFs and wish more were available.

Jailbreak was clearly still in development when I played and bugs were being discovered regularly (it is software, after all). I assume the architecture of the backend was designed to scale in such a way that maybe was never exploited.

Once the core of the game world is complete, how much effort is it to construct new content (new levels, for example)? I imagine the creative effort is more significant than the concrete construction of a level. I'm trying to understand if there is a way to balance the small subset of people who would pay for a "season" of a game like this. Is there a way to balance the effort and the price for such a thing to break even?

I'd pay to play a game polished like microcorruption (if you add an API documented like jailbreak). Especially if it were wrapped in a campaign/story. Combining open ended development/coding of such a task, the puzzle aspects of most exploit development, and a genre-fiction story would be a win for me.

I worry there aren't NEARLY enough people interested in such a thing but I wish there were.

For those of us who don't know what this was, what was it?

A startup by patio11 and tptacek to attempt to help recruit software engineers better.


The launch announcement: http://www.kalzumeus.com/2015/03/09/announcing-starfighter/

Launch announcement HN thread: https://news.ycombinator.com/item?id=9173939

Hats off to patio11 and tptacek; trying is far more than most people do.

I am not done. We're just done with this specific iteration.

I didn't think that my comment implied that you were giving up.

Sorry! I didn't mean to suggest you did. Just saying: it's not like a "I'm doing this in the long term" kind of thing. The next thing will be in the writeup I do for Starfighter.

The name reference is to "The Last Starfighter", an old movie wherein some aliens used a cabinet video game to find the humans with the greatest natural skill at piloting their space-combat vessels. If you win the video game, a robot is dispatched to the cabinet's location to recruit the player.

Naturally, Starfighter extended the premise that analysis of one's play can predict one's work performance to recruiting software professionals. Win the game; get hired.

I feel like they never explained what the name was in reference to. I had no idea why they were called that, especially when their only game was something to do with stocks.

> especially when their only game was something to do with stocks.

One of the two games/level sets was about stock market. The other one was about jailbreaking/reverse engineering.

A company that created capture the flag type games to be used for recruiting and hiring. See stockfighter.io for an idea.

Reference images for the upcoming enormous patio11 writeup: http://www.imdb.com/title/tt0087597/

I'd never seen that twitter account before. Not my cup of tea.

Things that seemed cool until I looked (even a little) closer at the people behind them: urbit, stockfighter, and ESR's homepage.

Urbit: Curtis Yarvin.

Stockfighter: Patrick McKenzie (patio11 on HN), Thomas Ptacek (tptacek on HN), and Erin Ptacek.

ESR's homepage: Eric Raymond.

I can see why someone might find something from Yarvin or Raymond less cool after looking (even a little) at Yarvin and Raymond because they have some controversial opinions outside of tech, but I've never heard any such things about the Stockfighter people. In fact, two of the Stockfighter people have the two highest karma accounts on HN, indicating that generally their opinions line up pretty will with the majority of HN. What did you find uncool about them?

Huh, I was under the impression that Starfighter never left beta.

Thanks for trying, guys. It was great fun.

I never used the service but I figured it'd be a strong effort given the folks behind it. Not knowing any operational or competitive details about the business, is this any kind of an indicator in lessened demand (or friction) for hiring devs?

Maybe twitter should hire tptacek's company for a security eval.

Applications are open for YC Summer 2019

Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | Legal | Apply to YC | Contact