Hacker Newsnew | past | comments | ask | show | jobs | submitlogin
Chatroulette Creator Is A 17-Year-Old Russian Kid (nytimes.com)
193 points by inmygarage on Feb 13, 2010 | hide | past | favorite | 96 comments


"I am glad that my relatives helped me with it by ‘investing’ some money in my idea. It wasn’t very much money, so I couldn’t just buy new servers just like that, I had to optimize my code as much as possible instead."

No choice but to optimize. Love it.


I have a question for Hacker News. I started working on a website extremely similar to Chatroulette about three months ago (with a few interesting differences). But I haven't finished it yet because I'm busy on other things in my life, I chose some wrong technologies to get started with, and I'm not the best developer in the world. Now that I've heard of Chatroulette, should I keep on going 'til I finish my project, or move on to the next thing?

My biggest concern is that I thought the site would be so cool that traffic would grow exponentially from organic word of mouth. Now that there is a Chatroulette, I'm afraid it loses its wow factor.

In fact, I can still think of some features that would lower the penis count and increase the usefulness/fun of the site.

edit: Is this query too off-topic? Should I start another thread? Seemed like a good place to ask.


You want an honest answer, Jacob? You're competing with a 17-year-old kid who made this web site for fun. I think you're taking yourself a bit seriously, and that's not a good thing.

Are you having fun with making it? Make it, then. Make a damn good web site. If it's damn good, people will use it. You can make this a damn good web site, right? It's not going to be a shitpile? I'm asking seriously, not rhetorically. Generally quality is more important than anything; if you've got it, you're set.

If you're not making it for fun, then frankly, I don't know why you'd have bothered to start in the first place. You don't think you're going to get rich off something like this, I hope. If you're learning something by making this site, you can still learn by making.

Or did you do it just to see the results? Because you can see the results on ChatRoulette. If you think you can do better, stop asking us if a better product will be worth anything and start asking yourself if you can really do better. If it's better, in a way that people care about, they'll use it. If it's not, then nobody will even care enough to mock you.

Are you doing something meaningful with your time? Answer honestly. If you are, then I can't see why you'd ask us anything of this sort. If you're not, then get yourself together and do cool shit.

(Somebody who uses this site more translate what I said into Hacker Newspeak. Something something penetration adaptability futureexperience something tippingpoint gladwell rand.)


Thanks for your reply. You have a clarity of thought in your comment that is actually very helpful. I feel a bit foolish asking dumb questions sometimes, but it gives me an opportunity to learn from people with better thinking on a subject.


Its not a dumb question, and I hope you haven't been discouraged from future participation. The advice you've been given here is excellent though. If I could add one thing, it would be this:

In your first paragraph, you've already made 3 excuses about why you haven't released anything:

   - "I'm busy on other things in my life"
   - "I chose some wrong technologies to get started with"
   - "I'm not the best developer in the world"
You've got to cut that shit out right now. Let this be the last time you got jumped on an idea because somebody else was working his or her ass off on it, while you were making excuses.


This advice is given on HN every day, guess it's my turn.

- Competitors are validation of the idea.

- The only way sites get the "wow" factor is through rapid iteration and contact with real users.


I guess I don't pay attention when other people ask the question. Thanks for putting up with my repetition.


Last year, me and a friend built a chat application which would connect you to someone random. We had an impressive backend, and some really neat features including mining abuse reports to extract offensive conversations, etc. (Even considered applying to YC with this)

A week before we were set to launch, omegle.com came out which was the same thing - simpler - but essentially the same. We abandoned the idea immediately because the internet was abuzz with omegle's brilliant concept, and we'd be seen as imitators and won't get enough people interested.

One week, that's all we lost by. If I hadn't pushed for combing the site for security holes and writing in spam protection, over-optimizing the comet-server, etc, it would've been our site over there getting several hundred thousands of hits. This taught me the importance of releasing early and iterating and improving quickly.

Still, every time I see omegle.com in the news or other sites, I feel sad that it could have been us, but I also feel proud that I came up with, and implemented an idea which would've worked nevertheless.


This is badly, badly, mistaken.

(a) You don't get buzz just by getting something out there. Usually, nobody cares when you launch, no matter how cool your idea is. What you were seeing was either effective PR or dumb luck, neither of which can be duplicated trivially.

(b) One friggin week advantage is nothing. Usually the first player in a space takes it in the shins. It's so exceedingly rare for a first mover to win that calling it an advantage seems to ignore market realities.


What a waste man, I've never even heard of omegle.

Keep that in mind please. In spite of whatever the buzz is today, except for the top 10 sites or so the majority of websites have not been visited yet by the majority of the people on the planet.

There definitely is room for more than one in almost every space except for social networks that are 'feature comparable', sooner or later one will gain the momentum and crowd the other one out.


Also, remember that you are _way more aware_ of your space than the typical person. Every day you are reading as much as you can in your area, so the chance of stumbling on a competitor is much larger for you than for your potential customers.

As a startup, don't assume that all your customers just read the great PR on your competitor. Most of them probably don't know that they want random chat (or live power monitoring)


I'm fond of the saying... the pioneers get the arrows, the settlers get the land.


Thanks for your advice. I never saw it this way.

Maybe I didn't give enough details in my original post. This was a little hack that I thought would be cool, and would help me learn how to write comet-based apps (which it did). When Omegle came out, was a little annoyed but since our project was a few days of work, I put the code away and focused on my main project. I didn't even hear of omegle again after then until a few weeks after that.

But still, my current project is chasing a first mover with a few months' head start and it's nice to get reaffirmation that we don't need to be first movers to get ahead.

P.S. If you're interested, here are slides from a presentation we gave in our local startup interaction session: http://www.slideshare.net/guest3cff7f/connr


“We abandoned the idea immediately because the internet was abuzz with omegle’s”

I can’t help but read that as giving up at the first sign of trouble. Sure, competition sucks, but if you had just a single advantage you could have used the Omegle buzz to your advantage and effectively ride their coattail, what many tablets are presently trying to do with the buzz surrounding the iPad.


Our tool wasn't a startup, it was a little app I wrote in a short while to learn how to work with comet servers. I don't subscribe to the "giving up is cowardice" line of thought. We had nothing really new to offer the world and I had a bigger, more important project to focus on.

While I will gladly compete when I feel that there is a potential for becoming better than the competitor (like I'm doing right now with my current startup), I can't bring myself to do something that offers nothing new and attempt to "out-market" the competition.


Out-marketing is where all the battles are won. You have to make people aware of what you're doing, not sit there and wait for them to do it.


Second comers to markets tend to do very well, since they can learn from the mistakes and the experiences of the innovator. Apple was hardly the first company to make an MP3 player, after all..

Look at fads in music for confirmation of this. Lady Gaga got big last year.. and now we have Ke$ha basically playing a similar card. Boy bands? Crap, when they come out, you end up with 101 boy band clones within a couple of years.

Get in on the fad, but execute in a way that makes you distinctive.


Yeah, I think it would be more fun if you could limit it to a community. Say, people that come from the same referrer get joined to the same "room". Then people on hacker news could have a link and we could all talk to each other randomly, and other communities could do the same. Would probably get much fewer penises.


I like your idea, but having spent a couple of days on ChatRoulette (and having seen the lack of success Seesmic has had in the blog comments sphere) I'm wondering whether there's really an appetite for video amongst strangers beyond seeing crazy/sick stuff or a gutter level of curiosity.

Even if you play ChatRoulette legitimately, the number of people who will actually talk back to you is ridiculously low. You can get into typed conversations with people and they won't open their mouths a bit. People also seem OK with just staring at each other. It's like an awkward subway ride.


http://meet.tinychat.com/put-anything-here (check back in a few hours)

Using nothing but the tinychat API to do it.


go for it...there'll always be competition, and Chat Roulette doesn't have the market penetration yet.

Just figure out a way to differentiate yourself.

1. Better layout, make it a truly glorious experience...a few extra KBs to load an interesting background is nothing when you are streaming gigs of data.

2. Focus on the U.S. factor. That guy is trying to have the best of both worlds, but Germany is really not that close. So focus your service on U.S. and Canada and offer faster speeds.

3. Offer easy ways to share screen caps. I dunno something like a report where you show screen cap of person #1 and screen cap of person #2 + the actual text. Host it on your own server, so that people(reddit/digg) can just link to the chats in question.


4. Some way, some how, get rid of all the dicks. Or at least try to make it opt-in, or figure out a way to ban/segregate repeat offenders.

Seriously. Chatroulette may be a great idea, but when 20% of the users are trying to shock the rest of them (or are "legitimately" perving out on camera), it destroys the return visitor rate. I know plenty of people, including myself, that were initially interested in the site, but couldn't get past the gross factor.


The best way to do well with any "2nd" product is to fix important problems in the first product.

ChatRoulette is not short of problems. Some people on reddit did a "study" where they found that some large percentage of "chats" were with masturbating men. Can you fix this? Probably. Add a check box saying 'show masturbating men' (I'm gonna say this should default to 'no'. Then when you end a conversation, allow a user to specific that they ended it because there was a masturbating man at the other end.

As far as I'm concerned, ChatRoulette is unusable until this is solved.


edit: Is this query too off-topic?

Obviously not. You seem to have an issue with self-validation. The most salient fact about chatroullette is that it is a piece of garbage. Find out why that is, then fix it, and put out your own service.

Quick idea: Have "categories", like "weight lifting" or "chess playing" or "NASCAR fans", etc, etc. And then have timed quizzes to prove that people actually fit the category. Then they can talk to similar enthusiasts.


...the simultaneous user count grew from 10 to 50, then from 50 to 100 and so on. Each time the user count grew, I had to rewrite my code completely, because my software and hardware couldn’t handle it all.

A nice real-world example of how getting something out there is more important than trying to achieve technical perfection first. He validated the concept, then worked out the technical issues as he scaled up. Meanwhile he had some revenue coming in to help fund development and hosting.


This has been the fastest growing site I've seen in a while.

One interesting sidenote: the word "roulette" is terribly hard to spell. Pretty much every typo domain for it got snatched up pretty quickly, and are currently showing AdSense for domains. (chatroullette.com, chatroulete.com, etc.)

Probably making a solid amount of money. I hope the owner himself is the one sitting on them.


It's only hard to spell if you have no clue about french ;-)

Here's a memory hook:

- "ou" because that's french for "oo", as in: douche.

- "l" since it comes from "roule" or "rouler", as in: joule, boule (there is no "ll" between "ou" and "e" in french, ever).

- "tt" since it's a female noun, as in: cassette, baguette, couchette.


Well I think most visitors don't know French, as the article indicates most are from the US. Even I misspelled the domain on my first try. doh!


Honestly, it took me a minute to realize that the overlap between French and English is actually smaller than I thought (and much smaller than between English and German). I had a hard time coming up with sticky analogies for the spelling of "roulette".


That would be an interesting advertising model.

Keep the main application advertisement free, but buy up all the doppelgangers and advertise on those.

Diabolical.


And how exactly would that work?

Every time a user, or potential user, mistypes your URL then best case they end up on some keyword related site. Worst case - they end up visiting some adword-buying competitor.

If it's more profitable to direct users away from your site than to visit there's a problem with your business model.


As a counterpoint to the dumb users googling for "facebook login" and typing their password into the site that was the first hit, here's a case where the people who get to chatroulette by googling for it come out ahead, because Google corrects the obvious typos and brings up the real site as the first result. ;-)


I assume for a Russian, roulette is easier to spell. :-)


made my day! P.S.: I'm Russian :D


Probably making a solid amount of money.

Having shamefully owned a few typo domains in the past, I can confirm this is true. The ad click-through rate on typo domains is absurdly high (~33% IME), probably because the ad is for the domain they intended to visit.


The site is also accessible at chatrt.com.


After reading this, I gave the site a try. It sucked. For half an hour or so, one person after another just clicked next. In that space, there was one person who asked "where u from?" and 2 doodlers. There were at least a dozen people filming themselves masturbating. Using the site was simultaneously revolting and mind-numbing.

It would be difficult to get a worse user experience from a website.


It doesn't take long to realise that truly random is not what you want - you want somewhat filtered so people who are being disgusting / absurd can do it in their own space.


countdown to the mixergy guy tracking him down. "HOW SOME RUSSIAN KID BUILT A SITE THAT'S BIGGER THAN YOURS" ...and how you could do it, too!


Frankly, I'm a little slow. I should have jumped on this the minute Fred Wilson brought up the site.

Update: Thanks for the prod. I sent him an email.


He is an awesome and inspiring guy! I really appreciate how honest he is still, and how not sure what the service actually is.


--I actually think that it would be best to found Chatroulette as a U.S.-based company. But this is just an idea.

If that thinking starts to change it will be devastating to the US business sector.


7 Gigabits/second...can someone translate that into a dollar amount for me?


7 Gb/s * 0.125GB/Gb * 86,400s/d * 30d/mo * 0.0009765625TB/GB = 2215TB/month to 738TB/month ("20,000 users at night", 1/3 peak usage - 8hrs)

> That’s why I have put only four links on the bottom as advertisements. And what is interesting, is that these advertisements almost cover all expenses, just those four links on the bottom!

EDIT: I give up on the conversion

I went to a talk once with Jeff Barr of Amazon Web Services. The owners of SmugMug charge all of their AWS on a corporate credit card. We're talking multiple roundtrip tickets from bonus points a month.


It's just over 200 terabytes.

http://www.google.com/search?q=0.7+gigabit%2Fsecond+*+86400+...

I've seen traffic advertised at €15/TB in Germany, that's €3.000/month.

EDIT: I screwed the units up a few times, sorry


You can simplify your query by letting Google do the units conversion. Eg, search for:

    86400 seconds/day + 1
To get the same result as your link google:

    0.7 gigabit/second * 30 days
Or even:

    0.7 gigabit/second * 1 month



7Gbits/s saturated over a month is 2,247 TB

http://www.google.com/search?hl=en&q=7Gb%2Fs+in+TB%2Fmon...

So at your €15/TB i would be €33,705 / month

Also €15/TB seems low as prices around the US are $0.10/GB so $100/TB which would equate to around $200k/month

Of course when you are pushing that much data you're going to want to pay per Mbit/s and with bandwidth as low as $3/Mbit we get around $21,000/month

What is probably happening here is that this guy has 7 dedicate servers at about $150 each per month that have a gigabit connection with say a couple of TB data allowance for each server and he's just saying that he has a capacity of 7Gb/s. The chat client is probably peer to peer so the video bandwidth for him is not that critical.


> 7Gbits/s saturated over a month is 2,247 TB

Even the best calculation is wrong when you plug in the wrong friggin data... I confused myself from converting gbit to GB and ended up with 0.7 gbit instead of 7 .. :-S

> Also €15/TB seems low

That's why I do business with them :) It's hetzner.de, and their "rootserver" (€49/month and up) comes with 2.000 GB free traffic. Extra is free, but capped at 10 mbit, or €14.90/TB. The servers are connected at 100 mbit, though. I'd guess that's just to keep the ethernet infrastructure cheap, and a gbit connection would be possible for a fee.

I'm not affiliated with them, just a happy customer.


This number is bogus. Why?

He uses consumer servers. Each server has a maximum of one gigabit ethernet connect. Like providers counting all their network interfaces together to have a high marketing number like "12321 Gbit/s backbone" that's just a number.

See my upcoming comment in the NYT blog about that.


I am not sure the TB/month figure quoted by others is a price that would scale to this level. As a new start-up without the ability/desire to commit for a year or any credit rating, you are likely to be paying about $6 per Mbps per month. As a reference for this price in July 2008, Cogent took the bottom out of the transit bandwidth market by offering $7 per Mbps (dropping to $6 for 1Gbps committ) - (see http://connectedplanetonline.com/mag/telecom_cogent_throws_d...). Prices may have dropped a bit in the last 15 months, but not by that much.

So at 7000 Mbps/second - I'd put the price at more like $42k per month.


Is it not possible to have flash directly connect with another flash user instead of routing all the traffic through central servers?


NAT/Firewalls are a bitch. I'm not a Flash expert, but as far as I know it can only open TCP connections, and only to specially set up servers (which respond positively to a policy request on a certain port). Punching holes in firewalls is pretty much a UDP-only thing in practice; Skype is one of the few apps that truly pull it off, and even that's not 100% reliable.

You could in theory do it with a signed Java applet, though it will ask the user for permission for access to his entire computer, and it's not something you'd throw together in a few weeks.


It looks like Flash 10.1 added something built in to do the UDP-firewall-punching and connection setup for peer-to-peer video between Flash clients: http://labs.adobe.com/technologies/stratus/

No idea if that's what he's using or he rolled his own. Using Stratus seems like the most obvious option, but it looks like it's limited beta only, and the beta EULA prohibits commercial use, so if he's using it, it's unofficially.


From my experience a couple of years ago, no, not without installing an extension to flash.


The cheapest bandwidth most can buy is about 2-3$ per mb/s in the US. Likely to be more expensive in europe.

That would mean each gigabit would cost between 2k-3k USD.


That adds up to around $5 billion a month. I think your estimate is a bit off ;-)


He probably means "2-3$ per megabit/sec".

So $14,000 - $21,000 per month. Still seems too much.


You can get a TB traffic for 15€ (Gigabit port) in Europe.


I don't know anything about streaming video - anyone got any idea what kind of software stack he'd be running for this, and what kind of custom development he would have had to do?


Flash + Flash Media Server (or clone, like Red5). It's just a matter of plugging together the pieces.


Is there a way to send the video stream directly to the other user, not through the server? Sort of P2P-like?


Yes, and this is what chatroulette does.

I did a TCP dump of the traffic and the data sent to the central servers is very minimal, mostly short plain-text messages containing little bits of information - your partner's camera status, number of users online, next button pressed, etc.


If thats the case, and its using stratus:

1: He is either misquoted or lieing about bandwidth usage in the article.

2: He is breaking adobe's tos for stratus, as you are not allowed to run ads along side any tech using it.


http://labs.adobe.com/technologies/eula/stratus.html

I don't see anything in the TOS against advertising.

Nor in the FAQ, with regards to commercial applications: http://labs.adobe.com/technologies/stratus/

Edit: That's an article. We're talking about the actual TOS, which does not specifically disallow advertising: i think they simply want to make sure people aren't billing for access to p2p services (which is stated clearly in the TOS).


Look harder then:

http://www.adobe.com/devnet/flashmediaserver/articles/p2p_ap...

"Stratus is for non-commercial use only."

Additionally, when you receive your beta access they show additional docs letting you know that if you want to run ads/make money you need to look into livecycle. Even had a adobe rep contact us after we launched one of our p2p services informing us of such.


I did not know about stratus (http://labs.adobe.com/technologies/stratus/), looks interesting! Now let's hope Red5 is planning to provide stratus/RTMFP support soon...


Are you absolutely sure about that? I don't think Flash supports that natively, so it would involve some real weird hackery to do something like run an RTMP server on both of the client machines in a Java applet or something, plus NAT traversal hole punching and other disgusting hackery.


I'm only sure because I saw a bunch of RTMP and Flex messages going to the central servers (they all have similar IP addresses) and a lot of UDP traffic going to a pretty much random IP address that changed with every person I talked to.

Also, Fogbugz copilot does something similar. It's described at a high level here: https://www.copilot.com/tech/ which links to http://nutss.gforge.cis.cornell.edu/stunt.php


So here's how it works: Flash 10 has built in p2p support. Firewalls can block this, so if the firewall to the far peer is blocked,the near peer falls back to Stratus (adobe Server), which probably has a socket? open to the far peer.


> Flash 10 has built in p2p support.

Oh!


Don't worry, so does Chromium.


If you create a standalone client yes. Otherwise domain policies and firewalls will prevent this.


Some things I find odd:

A) There is something different about this site. In a craigslist (no high production value type of way). It also has very strong network effects because as more people join, the whole site becomes even more valuable.

B) I feel like I need a friend with me as we play. It's odd that it feels like going into the city or a bar (never know what I'll find), but I'm also totally safe.

C) This kid ABSOLUTELY needs to meet up with PG, Craig Newmark, or somebody before he's taken advantage of by a crazy VC.


Quick, someone apply this format to dating!


already been done, search for internet speed dating or something along those lines .


woome.com


www.winkvid.com


Hes not even old enough to view the most of the cams


Why, does he have no eyes? You're using American cultural assumptions where they don't apply :-)


You are sooo cool---Chat Roulette Russian kid --- whoever you are. You are in fact a global peace maker and connector of people-ideas. I was a little revolted of course--like most---at first-but then I started playing my guitar for people and singing and they played back at me and we did songs together. But then of course it all broke down when one of the guys at the singing chat party ran into his bedroom---got a guitar and whipped of his clothes and put a sock on his dick---but you know---it's the thought that counts! kudos to chat roulette my friend from katy in eugene oregon!!


I am trying to learn German and I would love to chat with people from Germany in German. Is there a service like that?

I wonder if he could modify the randomness to at least just a single country.


Check out livemocha.com


Thanks!


There are language services like this that utilize skype or video chat. Names of which I can't remember.

My friend used to tutor Korean students in English over skype for some company.


Reminded of when voice chat was first available on win95 via netmeeting back in the day where there were maybe 50 people tinkering with it and you just connect to any of the other early adopting nerds to tell each other how frikin cool it was. Random, anon, but friendly.


Best chat roulette screen shots: http://www.buzzfeed.com/awesomer/the-24-best-chat-roulette-s...

I kind of liked the jolly nazi&israeli couple doing thumbs up. :-)


honeslty...an amazing site would be one were you can do the same video chat as on chatroulette but u could pick at what u are looking for....such a girls/guys who just want to talk ... girls/guys who want some webcaming pleasure...

that would be the best site :0 thx


That's the Moscow State University in the photo's background. Suddenly I feel so warm!


From the article: "I would love to visit the United States."

If people like him wanted so hard to travel to my country, I'd already be rallying my government to give those people a damn grant. Time for the USA to reconsider whether the TSA's "security" pays.


What? What does 'TSA security' have to do with some kid saying he wants to visit the US? Can't believe this comment was upvoted.


I think you misunderestimate the animosity many people hold towards the TSA.

The upvotes likely reflect that animosity as much or more than simple agreement with the implication that there's some sort of connection between the TSA and the kid's inability to travel to the US. Not to mention the rest of the comment regarding giving the kid a grant.


Except, really, it's more likely to be because Russia is not in the Visa Waiver Program that makes it super easy for most Europeans to head to the US on a whim.


He'd be a lot easier to pack as a carry-on without the xray machines


A 17-Year-Old Russian Kid

As opposed to a 17 year old Russian adult? Or a 17 year old Russian senior citizen?


moot for the masses.




Consider applying for YC's Fall 2025 batch! Applications are open till Aug 4

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

Search: