Adam Smith

05/31/2010

The Upcoming Generation of Lossless Data Compression, Part 1

First, yes – there is such a thing as next gen lossless data compression.

Here’s one for starters: cross-file content-aware dedup for images and video.

Take a JPEG image, compress it with your favorite lossless compression algorithm, say BZip2 or LZMA, and you’ll get a larger file. Of course that makes sense, as JPEG itself is written to be compressed out of the box.

Now take a group of, say, 100 JPEGs, and compress them into a single archive. You’ll get no real compression. Even though there are likely redundancies among those photos, JPEG isn't designed to handle cross file compression, and LZMA can't handle it because the redundancies aren't obvious by looking only at the binary patterns.

But do that with a compression algorithm from the future, and you’ll get a much smaller file. These algorithms will know how to do content-aware cross-file compression.

What would these algorithms look like?

In fact we already have algorithms that know how to de-dup among a set of pictures. They’re called video compression algos. Try taking those same 100 JPEGs, put them into a movie file format, and run your favorite video compression algorithm over them with the right parameters and you’ll get your smaller compressed archive. Of course this is just a hacked approach, but you get the picture!

There are also various tricks you could do on text, audio, etc.

This technology will become increasingly important as more and more data gets centralized into the cloud.

Someone from the commercial side is already working on this. After thinking about this idea for several hours and getting really excited, I came across a company called Ocarina (chalk talk here) who is commercializing this approach.

That doesn’t mean this technology should be limited to big companies with big budgets, though. I’m hoping someone from academia or the open source community will eventually pick this up and run with it!

Part 2 to come soon! (Edit: link to part 2) In the mean time let’s hear it if you have any further thoughts or ideas!

(There was a good discussion of this blog post on Hacker News here.)

05/04/2010

From Zero to a Million Users - Dropbox and Xobni lessons learned

I just finished a talk at Web 2.0 Expo titled From Zero to a Million Users with Drew Houston, the founder of Dropbox.  It's about what we wish we had known years ago when it comes to growing your user base from 1 to 1M users.  We had a lot of fun with it.  The slides are below!

03/22/2010

Amazon S3's Pricing Model is Arbitragable, and the Future of Cloud Storage

Here’s what you do.

Amazon S3 charges money for each gigabyte that you store. You go to their top 50 customers, and tell them: we’ll dedup your data with the other top customers, resulting in overall cost savings. We’ll split those cost savings with you 50/50.

You're basically stealing the economies of scale of redundancy elimination from Amazon.

You could also get sophisticated with delta encoding blocks of data that aren’t duplicates but are close. Here’s a paper that goes detail on redundancy elimination.

You’d have to get people to trust you to not lose their data. Maybe you use best practices, hire cperciva to write it, and get an insurance company to provide a $100 M policy against data loss that’s your fault – but it’d still be hard to pull off.

This line of reasoning raises questions about the future of cloud storage. Mostly: does broad scale redundancy elimination matter?

If it does matter, we will see economies of scale in cloud storage. Does that mean there will be one or only a few major providers, or will redundancy elimination be provided as a layer on top that’s operated by someone else? Will cloud storage providers be forced to change their pricing models?

These are the questions that make our field so exciting!

What do you think: will large scale redundancy elimination matter?

03/17/2010

The Market Opportunity to Undercut Sonos

New technologies often start out serving the high end of the market where quality is more important than price, and over time a new disruptor comes in to serve the lower end of the market. This is a classic part of the innovator’s dilemma.

It’s also pretty universally understood that geeks love their Sonos sound systems. These boxes of magic synchronize music playback across speakers spread around your home, all wirelessly. For example, right now I’m playing music from my PC into a Sonos line-in that gets piped into my living room speakers. If I walk upstairs to do laundry I can pull out my iphone and use the remote control app to route that music into my upstairs speakers as well. Playback is synchronized.

A low end Sonos system costs $800.

They also have all kinds of fancy features like integration/partnerships with Rhapsody, Napster, Audible, Sirius Radio, etc etc. Their boxes have optical audio ports, ethernet ports, etc. None of these things will matter on the low end.

So what, exactly, would the low end solution look like? Easy: small wall-warts that have a wifi chip, an audio card (D2A and A2D chips), and a microcontroller/DSP to make it all work. On the outside it would have two audio jacks: one for audio in and one for audio out.

(You might also sell a set of speakers with your device built in, since the less hacker-ish part of the market will want a total solution.)

Now, I have zero experience with hardware but I’d guess you could get the bill of materials down to $40, without speakers. If you are super scrappy you should be able to bring the initial product to market for $750k. But I could be totally off base.

The goal would be a cost point that's ten times cheaper.

Here’s the coolest part. Using Sonos’ audio streaming today requires two Sonos boxes – one as the audio source and the other as the sink. A standalone computer can’t be a source; if you want to play music from your laptop and move around you’re hosed. (There is a way to do this, but you have to set up and use their player software, which is not the right solution.)

The low end disruptor would need this “virtual sound card” feature, for both PC and Mac OS. The lowest common denominator would be having one of your devices plugged into your stereo system and you can stream music to it from your laptop. And people can upgrade over time by just buying more of your cheap devices.

The only other meaningful player in this space is Apple’s AirPort Express, which can stream itunes to an audio jack. I don’t believe you can play from any audio source on the computer, and I don’t believe you can stream to multiple AirPorts at the same time. Update, 2013: Apple has significantly improved AirPort streaming since this original article.

It's usually not a good idea to compete on price, which is why I'm less interested in this idea. There certainly seems to be a "there" there, however.

02/16/2010

The Great Q&A Wars of 2009 ~ 2014

With Aardvark’s sale to Google last week and StackOverflow’s announcement yesterday that they are going to raise VC funding, we close the first chapter in the Great Q&A Wars of 2009-2014.

The major players are now Quora, StackOverflow, and Hunch. Not to mention incumbents like Yahoo Answers, and likely a bunch of startups that haven't been started yet.

Who are these companies, and what will determine who wins, and who loses?

Quora was started by Adam D’Angelo, who was a cofounder at Facebook, and their CTO. Now that’s horsepower. You can see the Facebook DNA in Quora’s product. It has a news feed, profile pictures abound, and there’s a strong emphasis on ‘following’ people and topics. In round one of the wars Quora amassed a small user base of Silicon Valley who’s-who.

Hunch was started by Tom Pinckney and Chris Dixon, two heavyweights who started and sold siteadvisor to McAfee in ‘06. I was consulting at siteadvisor when they sold. These guys are pros. They recruited Caterina Fake almost two years ago and recently got Jimmy Wales, Wikipedia founder, to join their board.

Hunch doesn’t have a Q&A format but I include them because they have the same ambition of facilitating social knowledge exchange.

You can see how it works at their page of hangover cures. There’s a whole list of cures but answering questions like “Can you keep anything down?” gives more specific suggestions. It’s similar to decision trees.

With that background in mind, here are some questions that will be important in the upcoming years:
 

1. How much data will the winner need about its users? User data can be used to suggest answers and questions that you might be interested in.
   
   Aardvark had about 40 bytes of information about me. They knew I was into startups, programming, and San Francisco. That didn’t seem enough to direct questions to the best person.  I regularly got questions I wasn’t interested in.
   
   Hunch amasses profile data by giving you the opportunity to answer personality and interest questions.
   
   Quora makes user profiles more of a first class citizen. They will be able to use data from Facebook Connect, and should be able to make inferences based on who you follow / who follows you.
   
   It’s not clear (a) how important it is to know a lot about your users, and (b) if so to what scale?
   
   For example crawling your email could give a website kilobytes worth of valuable data. Imagine if Hunch had your Amazon purchase history!
   
   StackOverflow so far lacks a strong strategy for building user profiles.
   
   (I would have liked a downloadable tool that would svn blame all of my code to figure out what I’m an expert in and what I’m interested in. But no dice, and this approach certainly wouldn’t work outside of StackOverflow’s current focus on programming topics.)
   
   
   
2. Free text versus structured. Hunch’s data is heavily structured. Quora and StackOverflow are largely free-text.
   
   The structure certainly helps for questions like What digital camera should I buy?
   
   Quora and SO would choke on these questions in their broad form. Hunch takes you through a nice decision tree that asks about your price range, mobility needs, etc.
   
   It’s not yet entirely clear how important this type of structure is in the product.
   
   
   
3. Managing the signal-to-noise ratio. None of these three services have hit mainstream in the way Yahoo Answers has. God bless the Yahoo Answers team. Just take a look at the relationships category and you’ll know what I mean.
   
   This might or might not turn into a significant problem. If so, Quora’s use of the social graph will be a robust solution.
   
   
   
4. Monetization. To the extent one should worry about monetization this early in the game, Hunch seems to have been thinking about it more. Several of Chris Dixon’s (CEO) blog posts (example) have been about monetizing intent.  Selecting a camera to buy is definitely a compelling example.
   
   Quora seems to have a more Facebook-like approach to monetization, i.e. plan for lots of scale and then add advertising.  It is so early in the game though that this is a random guess.
   
   
   
5. Site organization and browsing. Yahoo Answers uses categories and time sorting within each category. Hunch seems to surface popular and interesting questions on their home page.  Quora has ‘following’ and a social graph approach. StackOverflow uses tags and ‘sub-reddit’ like sites such as answers.onstartups.com.
   
   Ironically none of these approaches are the same and if you gave me a list of approaches and a list of companies I probably could have predicted who used what.
   
   I'd guess there's something to be said for a blend of the Quora and Hunch approaches. Some problems are local to your social graph, and others are universal problems…like hangovers.
   
   I don't think any site has the final answer on this question yet.
   
   
   
6. Leadership.  All of these teams have great people, hands down.
   
   That said, StackOverflow’s announcement today seemed to suggest that they will be bringing in an outside CEO. Though I don’t know the behind-the-scene details, this seems like it will disadvantage them.
   
   Furthermore, it seems like “everyone on the SO team works remotely from home,” and a recent job posting suggests that might continue. I hope not.
   
   Frankly I’d like to see Jeff Atwood stay CEO and fully commit to StackOverflow in the “startup” sense of commit. That means dreaming about the company's problems at night, not talking at too many conferences, or doing other fake CEO stuff. 
   
   

Someone is going to win. We are at the Model T stage for knowledge sharing online. I’m totally pumped that these guys are working on this problem!!

What about you? What other open questions are there for chapter two in this market?

01/24/2010

Magic in the software -- what the point and shoot camera industry needs

Point and shoot cameras are being displaced by smartphone cameras, mainly because smartphones are more convenient.

But that isn't the whole story. In the coming years point and shoots will find themselves under attack from a different angle; software will be easier to write for smartphone cameras, enabling use cases that point and shoots will have to scramble to keep up with.

For example, I take pictures on my iphone using the Dropbox app. Pictures I take are immediately copied to all of my computers. Even if point and shoots had internet connections, which they don't, they also don't have app platforms for companies like Dropbox to come in and add value.

There are a ton of apps that remain out of reach to point and shoots. Just look at the variety of camera apps for the iphone. Many of those apps would make sense in the point and shoot world.

So here's what I would do if I were a major camera manufacturer: create an app store for consumer camera apps.

Make sure the cameras have large-ish LCD touchscreens. Maybe even add wifi adapters that can work in a peer-to-peer mode. Give the apps access to the SD card, let them remix photos, show UI, and even use the wifi to share photos with other nearby cameras.

The magic is in the software.

Unfortunately I don't see this happening. This is a classic innovator's dilemma situation. It will start on the low end and work its way up market.

01/13/2010

Seven Major Websites that Send Passwords Unprotected, and State Sponsored Deep Packet Inspection

I was having a conversation yesterday with my close friend and roommate Anson Tsai about websites that send user passwords in the clear. This only matters, in theory, if there's a man in the middle who can read your traffic.

So it was a huge coincidence that China was accused today of coordinating sophisticated cyberattacks in an attempt to access the GMail accounts of some Chinese human rights activists.

Putting two and two together, I started worrying about China's ability to harvest passwords from Chinese users using deep packet inspection at their Great Firewall.

And it's only going to get worse as states get more sophisticated.

So I went through 36 top web sites and sniffed my computer's network traffic while logging in to see if the fake password I entered was sent in plain sight.

Seven of the 36 sites I tested sent passwords in the clear, available for an Internet Service Provider to read. That's 20%!!

Here are the sites that send passwords in the clear:

• Slide
• Wikipedia
• hi5
• Photobucket
• Gamespot
• Tudou
• Taobao.com

Taobao.com and Tudou are Chinese. Interestingly, 50% of the Chinese websites I tested were offenders.

In ordinary contexts I wouldn't be worried about this, but because of some states' unique Internet regimes I'd guess that, to take an example, the Chinese government has passwords for at least half of its citizens.

Let's do our best to fix this situation. I don't know anyone at the Chinese sites but let's push our contacts at the other sites to fix their security bugs. There are well known, easily implementable techniques for securing passwords sent back to a server. It shouldn't be a big ask.

It will be difficult because the threats are not immediate or perceivable, but I think as top Internet properties these sites have a responsibility to use basic security practices to protect its users.

Expect a follow up report and please let us know in the comments if you can help or reach out to folks at Slide, Wikipedia, hi5, Photobucket, or Gamespot!

Feel free to retweet. Here's a link to this post you can share: http://bit.ly/8f4Wsl.

01/07/2010

Technology to circumvent online copyright enforcement

This post could also be called “Why it might become civil disobedience to serve up random data.”  You’ll see why in a bit!

Also, this post requires a basic understanding of XOR and how a one-time pad works.

If I serve up Lady GaGa’s latest MP3, I’ll get a takedown notice for copyright violation.

What if a friend and I served up bits that, when XOR’ed together, became the latest Lady GaGa MP3?

A judge would likely find both of us guilty of copyright violation, but it seems like a small stretch of the law. I wonder if there’s something there…

What if I served up bits that, when XOR’ed with some publicly well known bits (e.g. the Firefox installer), became a Lady GaGa MP3?

I would probably be liable; it’d be strange for Firefox to be found guilty in any way. You can see the law now: “you are culpable if you distribute something that is intended to be a reflection of a copyrighted work”. Base guilt on intent and let the court discern intent.

This approach seems reasonable until you imagine it at scale. Any given copyrighted work could be expressed across 10 random-looking files. Are the copyright owners prepared to sue all of them?  That’s denial of service for you.  But it gets better because some of the hosts might really be innocent; some of the data blocks might be legitimate, for example ciphertext that just looks like random data.

In other words, this system would either (a) circumvent copyright enforcement, or (b) allow me to get arbitrary hosts of random-looking data sued.

Hm. College kids could upload random data to their home directories, just to help out the effort.

I don't think this will ever happen, though it is a fun thought exercise.

Update: There has been a lot of great discussion on this over at Hacker News, in addition to the great comments we've been getting below!

Update 2: This idea isn't intended to indicate that file sharing can suddenly become legal if done in a certain way. It's intended to make enforcement intractable.

01/06/2010

How to Find and Hire Amazing People, Part 4

Hello folks!  This is the final part to the four part hiring series.  Parts one, two, and three are here, here, and here.  As the final part to this series, I want to close us out with a quick thought and links to other great startup hiring reading!

My closing thought is that there are different styles to everything, and your mileage may vary.  There's more than one way to skin a cat.  And so on.

Dharmesh likes to ask employees to work through a dating period of 60-90 days.  Jason Cohen likes to ask candidates to answer some essay questions before an interview.  These guys are both awesome.  You can't just write off these differences.

So you'll develop your own process over time that fits your style, industry, and company situation.

Which is great!  That's part of what makes your career fun and challenging!

Food For Thought

Here are some links to other pieces of writing on the subject of hiring.  I found all of these interesting and thought provoking!  Enjoy!

• Marc Andreesen: How to hire the best people you've ever worked with
• OnStartups.com:
• Why Startups Should ALWAYS Compromise When Hiring
  
• Startup Hiring: Why You Should Date Before Getting Married
• 5 Quick Pointers On Startup Hiring
 
• A Smart Bear: Hiring Employee #1
• Tom Pinckney: Competing with large companies for engineers
• Paul English: hiring religion
• SEOmozBlog: The It Getters

Thanks for reading, and stay tuned for future posts!

Finally, please let us know in the comments if you have any of your own favorite startup hiring reading to contribute to the pile!

12/14/2009

How to Find and Hire Amazing People, Part 3

This is part three of my series on hiring. The first part covered finding great candidates. The second part covered interviewing. So once you've found someone you'd like to recruit, how do you pull them in?

Recruiting

This is where you switch from "we're hiring" to "we're recruiting" mode. There's no longer any doubt about whether you want the person or not. Your goal is to get them on your team and make sure the relationship is set up for success.

And you better really decide if you want the person or not. The only acceptable result of interviewing for a hiring manager is PASS or RECRUIT. If you're unsure then do more interviewing, do reference checks, whatever you need to do. If you're still unsure then it's probably a PASS.

If you're uncertain then you'll fail at recruiting. They will sense your ambivalence.

One mode

There's only one mode of recruiting, and it's called 11. If you were 51% on the side of hiring a person, and you decide to hire them, you better go after them like they're Michael Jordan.

There are a couple of reasons for this. Most of it has to do with psychology, which I'll discuss more below, but the net result is you'll lose if you pursue your recruits with an effort that's proportional to their skill level.

And if you don't personally have the time or energy to recruit with vigor then you should find someone else to be the hiring manager.

Wooing and momentum

Once you know that you really want this person on your team, you can fully dedicate yourself to closing the deal.

Momentum is the name of the game. If you wait a week between the onsite interview and your follow up phone call your success rate will fall 5x. Either someone else will close them or they will reject you in their mind because they think you're going to reject them.

It's kind of like dating. There are the same "Will you like me? I like you" dynamics. It's a little less personal for you because you're just "the company" but it's more personal for them because they themselves are on trial.

This is why the best, happiest, most successful hiring deals you hear about happen in very small time frames.

So call them the day after the interview, and tell them that you'd like to work with them, and would they like to work with you? And you're willing to do anything you can to help them decide.

And it really is in your best interests to help them make the right decision for them. The only stable employer-employee relationship is a win-win relationship. If they won't be happy and fulfilled working for you then it'll become a net loss for you, too.

They might be worried about a number of things: how much responsibility will they have? How hard do you expect them to work? Will they be able to move up in the organization? Will they fit in with their coworkers? Etc.

These are all legitimate concerns. You should help them answer these questions as much as humanly possible.

They might also be worried about more career focused things like what their title will be, how much exposure to investors they will get, will they be reporting to someone high in the organization, etc.

Those are also legitimate concerns. They're just a little more charged and negotiating here probably has a lower real impact on the candidate's happiness than he or she thinks. The truth is that seniority is earned not taken. If someone negotiates ahead here they will rarely accrue real long term value as a result.

For example, someone out of college may negotiate to be the "CMO" of a startup, but the people whose opinions really matter (not Aunt Mae) know that titles mean nothing and real capabilities mean everything.

So these conversations are mostly misdirection. [1]

Permitting these conversations in your culture will also create added politics. Avoid it!

They might also ask about compensation at this point. Of course they need to know this before signing an offer letter, but it's worth working through the question of "If compensation were not an issue, or let's just assume 'market' salary, would you want to work with us?" first and foremost. That's the real meat of the question. (More on compensation below.)

Overpromising

Be careful about overpromising. You're most likely to overpromise on small or fuzzy things.

One of our early employees asked if he could get biannual reviews instead of annual reviews. Heck we had never done a "review" before; why not? It was a verbal thing that lasted thirty seconds. I found out a year later he was upset that we were just now doing his first review. Ouch.

We told one of our first employees that he would be like a cofounder or a VP. He later resented it when he wasn't included in various high level discussions. Triple ouch.

The truth is when a junior hiring manager is working with a junior recruit, BOTH of them want to be on the overpromising side. "Yes, I will work my ass off and I have a 200 IQ," says the junior candidate. "Yes, I will make you rich and happy every morning you come into the office," says the junior hiring manager.

Senior hiring managers frankly know better. Not only do they have a better sense of reality, they will also err on the side of under promising.

The next step

Once you and the recruit decide you want to work together, you've got to figure out the right compensation package. Then you'll have a verbal acceptance. Send over an offer letter that codifies the compensation and other terms you guys have been discussing. It will usually be two or three pages. Try not to make it sound lawyerific. Express your enthusiasm and make it upbeat!

It should include the start date you guys have agreed to.

Once it's signed you've got a non-binding agreement. (Work relationships are never binding.) You can expect the recruit to show up on the agreed date.

Now send them a nice care package with schwag. :)

How to woo

In the spirit of momentum and Andy Grove ("only the paranoid survive"), from the minute you meet them until 30 days after they start working for you, you want to be woo'ing.

Try to minimize the maximum amount of time that goes by without contact from you or your team.

Invent reasons to have touch points. A week before they start ask what keyboard and mouse they want. If you find the right moment and feel inspired, send off an email letting them know what's going on inside the company.

Have one of your board members call or email to help you recruit.

You want them to be excited and pumped!!!

Regular contact is most important while they're trying to decide if they want to join. An object at rest tends to stay at rest. If they have a burning question they might be too shy to give you a call. It's your responsibility to provide the regular contact. (Or, for the techies among us, the carrier signal.)

Also, have fun! New beginnings should always be fun!

Momentum, momentum. One speed only. Touch points, and have fun!

Compensation

I skipped over compensation earlier because I wanted to cover the high level arc first.

Now for a level of detail on compensation.

After you've decided you want to work together you have a good foundation to start the conversation about salary, equity, signing bonuses, etc.

Negotiation is something that every founder should read and learn about. If you have someone nearby (a friend, or investor) who has experience negotiating compensation, you might ask them to help you at first so you can learn from them.

Some top tier VC firms even have partners who are dedicated to helping their startups plan and execute their compensation plans. This can be helpful for both learning how to negotiate and to make sure your head is screwed on straight about the market rate for various positions.

Startups also go through different stages. My cofounder and I took an almost zero salary the first year, a $70k salary the next year, and up from there.

That's what founders do. Employees are rightfully less willing to take below market salaries.

If you ever find a startup where everyone is below market that means it's filled with people who are really hungry (good) and/or just out of college (comes with good and bad).

At the end of the day though, most startups just have to pay market salaries.

That's what you want to shoot for. Are you paying what's below market, at market, or above market? You want to be at market, for a variety of reasons.

So does the employee, in most cases, even if they don't know it.

Here's why. Compensation is just one of the many factors that people should think about when they're choosing where to work. Their team and their projects should be equally forceful, not to mention a million other smaller things like the commute, expected hours, job security, etc.

If you pay above market salaries you might be drowning all of the other factors out of the equation. That might work for finance or other job roles that are easily measurable or short term, but you want to build the team that will create the next great software product, and that takes creativity, teamwork, and dedication.

You also want someone who would want to work with you even if money didn't exist. Paying above market salaries is not a way to find such people.

There are ten other reasons why it's bad to pay above market compensation, other than just spending more money and equity than you need to. We'll stop here.

I'll also leave it as an exercise to the reader for why BELOW market salaries might create adverse selection.

So my whole strategy is to try to zero out the compensation piece of the puzzle and engage on all of the other levels that I think are more important. That's why I try to frame the initial question around "If compensation weren't an issue, would we want to work together?"

Note the market comp for managers or executives might be more than you want to pay if you're a first time entrepreneur. Especially if you are young. You are going to have sticker shock.

That's just because you have different life and risk equations. Those people are definitely worth their weight in gold if they are good. Without them you can't scale. And because they represent scale they get to capture some of that scale in their compensation.

They will still be a big win. Or a big loss. Managers and executives rarely have an ambiguous return on investment, so twenty or thirty thousand here or there won't swing the sign on that equation. Be willing to pay up.

Avoid Pressured Hires

Try to avoid getting in situations where you needed a hire yesterday. Urgency creates triage situations that will make you less scrupulous.

And god help you if you didn't have good hiring standards to begin with. They will fall even further.

If you think about it this has widespread implications.

It means a VC can kill a company by pressuring them to grow too fast.

It means the board and its CEO should be looking around corners to anticipate upcoming organizational stresses.

It means you should hire generalists early on so there is no need to hire fresh blood to tackle that new important problem. (You can always hire outsiders; you just don't want to NEED to.)

In general this comes down to "Be Happy." Of course you want to recruit in new rockstars whenever you have the chance. Of course there are roles you'd love to fill with the right person. But if it's the wrong person or the wrong terms you always want to be able to say "No thanks" without losing sleep.

A good clip

We were able to hire one high quality person per month at Xobni once we started hiring.

But it happened in bursts, so don't be worried if you don't hit your hiring goals for any single quarter. For example, we would buckle down for two or three months to push out a new release, followed by a lucky month of two or three new employees.

Be kicking ass

So now that you've come to essentially the end of this blog post series on hiring, I'll tell you the secret to recruiting.

Be doing very well.

For some reason nobody talks about this.

You can talk all you want about how to hire, how to find great people, and you can pound the pavement all day long, but you might be better off working on getting product traction instead.

The same secret applies to raising money as well. You can talk yourself blue about how to raise money, but the most significant bit will always be how much traction you're getting.

This is why Google was able to hire so many rockstars during their IPO days. And it's why facebook will be proving this rule again in the coming years.

I realize this creates a chicken and egg problem.

That's why technical skills are so important for founding teams. :)

And that's just the brutal truth of the situation.

Up next

Part four will be a bonus round. Much shorter. I promise.

Notes

[1] The truth is that perception does matter in some cases, but those betting on seeming impressive, rather than being impressive, will lose out on average. Thanks to Paul Graham for the distinction!

11/16/2009

My Startup Bootcamp Talk

The guys behind the MIT Startup Bootcamp have finished posting all of the day's talks online. They were nice enough to invite me to share my experiences and lessons learned starting Xobni at this event.

Overall I thought the talk went pretty well. The first half was slower paced than I would have liked. My favorite part were the questions! Maybe next time I'll just build the whole thing around questions, not sure.

BTW, the Startup Bootcamp organizers are awesome! They are "just" MIT students. Unpaid. On their own initiative and time. ...and they follow through over the subsequent weeks by formatting these videos brilliantly in HD.

Frankly, I'm hoping they come talk to us whenever they are considering career options!

Here's the video, below.

Here's Drew.

...and Dharmesh!

Thanks again to Michael Grinich and the other wonderful conference organizers!

11/11/2009

How To Find A Market For Your New Company, Family Edition

I was at a conference last weekend with a lot of non-tech entrepreneurs. It can be refreshing to get outside the valley and into a pool with different kinds of fish. It was a lot of fun.

My favorite question to ask new people was: How did you get into your business?

Now, these guys do things like importing soy sauce and related goods from Asia, or designing and manufacturing green buildings all across the US. It's really quite varied.

Most folks said they got into their industry because of someone in their family.

That's pretty cool if you sit down and think about it. It's a pretty robust way to allocate human capital in a distributed way. I doubt many journalists are telling their kids to go into journalism, but I'd bet most energy scientists are thinking that their offspring should consider following in their footsteps.

It seems slow as a mechanism of action, but probably only to those in highly volatile industries like high tech. On the grand scale of things it's probably about right in terms of balancing too fast with too slow.

And your family members can train you.

How many of your friends in real estate got into it because someone in their family showed them the ropes? Right.

Same thing with teaching. We have several public school teachers in my family. The ones in the business show the new entrants how to get certification, how to find a good job, what to expect, and so on.

My dad buys and sells used truck and heavy construction equipment. He got into the business thirty years ago because that's what my mom's dad was doing.

This is an amazing system!!

This doesn't seem to happen as much in high tech. We're certainly in a new field that hasn't had a lot of time to replicate through families. I know that Trip Adler, the founder of Scribd, was born to a doctor who started a company that later became public.

Topher Conway just started working with his dad, Ron Conway, who is the most prolific angel investor in the world.

So if you're looking for the next step in your career, look around you at your close friends and family. They're the ones who will give you the honest and informed skinny, and they'll be the most likely to help you out along the way.

How about you? Do you know anyone who has entered a career because of a family member or close friend? Let me know in the comments!

Editor's note: I still intend on wrapping up my series on hiring soon. Also, more ways MIT didn't prepare me for startups are on the way! Subscribe via RSS to stay in touch!

11/04/2009

How to Find and Hire Amazing People, Part 2

In part 1 of this series we talked about how to find people who might be rockstars. So how can you tell if someone is going to knock balls out of the park for you and your team?

Hard to say. :)

Here's the process we take a candidate through when hiring. Some of what I'm going to say is particular to interviewing developers, which is what I have the most experience with. Most will be general.

Resumes

First, read the resume.

The first thing I look for is experience at a software products company. So the company’s main gig has to be software, and their goods are sold as products not services/consulting. E.g. Microsoft, Amazon, Yahoo, etc. These are the companies that write production code as their core competency.

How long do they stay at jobs? Three or more years at a job is good. One to two is okay provided it's not the pattern. Less than one year is bad.

Next I look at their career trajectory, if they have one.

If they have less than three years of experience, look at their education. Small points for attending a top university for their field. Extra points if they seemed to really push themselves with grad classes, research, etc.

Lots of buzz words or acronyms in a resume is a bad sign.

It's great if they have a list of personal projects. It shows initiative and creativity. It's also good fodder for interviews down the road.

These are just some of my if-then rules. You'll end up growing your own set of rules over time.

Be very picky at this stage if the resume came from a job board or recruiter. You can read 100 resumes in the time it takes you to do a phone interview, so the cost of going to the next stage is quite high. Only do it for great candidates.

Resume Extra Credit

Smg on Hacker News suggested in response to part 1 that you can use online coding problems to help filter through the wave of resumes you'll get from job boards and optionally recruiters.

Justin.tv does this has some pretty neat problems on their jobs page.

Phone Interviews

40-60 minutes.

The most important thing is to have them write code or otherwise demonstrate what they'd actually be doing day to day.

Just to be clear, the most important part of ANY interview is where they demo what they'll be doing every day. It's not their resume, and it's not the stories about their old teams. It's actually asking them to write code, or actually asking a product manager to walk you through a product they've built.

You'd be surprised how many interviews I've been through where the candidate was doing wonderfully, right up to the skill demo where they bomb.

This doesn't mean you can start right off with the coding questions, though. You want the candidate to get comfortable first so they can show you their best colors.

Ergo I split my interviews into three parts: first, I have them walk me through their resume. Then we do "more interviewy type stuff," as I call it. And finally they get to ask me any questions they have about us and what we're up to.

I tell them upfront that's the game plan so they know what to expect.

Having them walk me through their resume helps them settle in. The most important things to dig into during the resume review stage are why they left each position, and try to triangulate what their manager thought of them. This should take about ten minutes.

Then comes the skill demo.

For developers I'll start with something easy like “Are you familiar with the singleton design pattern?" And then go right into my coding question. I have them write their code on a piece of paper, and have them read it back to me token by token when they're done. I copy it down on my end of the phone, and then we talk about it.

The code you have them write should be ten to twenty lines long. They can use any language of their choice, but it should compile. (For all you youngsters, "compile" is something that used to be done to code back in the day.)

It should be one function, with a loop, and have at least one or two edge cases.

It's amazing how much information such a short question can give you. Some candidates will nail the code in 30 seconds, others will take 15-25 minutes. There's at least an order of magnitude difference in the variation of outcomes for a simple 15 lines of code. That's awesome interview power and the most effective signal per minute part of the interview.

On the other hand I haven't found logic puzzles to be useful. I will follow up the coding question with some algorithmic questions if they're doing well.

Third part is where they get to ask you questions. Everyone they talk to should give them this opportunity at each interview. Usually doesn’t take more than 5 minutes.

Score the candidate.

One phone interview should be enough to decide whether to bring them on site, save for extraordinary circumstances.

Onsite Interviews

Three or four interviews, usually three. If an early interviewer gets negative feedback we send the candidate on their way so we don’t tell them up front what their interview schedule will be. Just ask them to be available all day.

Plan the interview loop out beforehand.

We also give the candidates a binder at the beginning with a picture and bio for everyone in the company. That's a great way to help them connect with and understand your team.

Have fun! Start with a big smile! Ask them if they'd like anything to drink. When you're done tell they can take a short break to stretch or use the bathroom. Being interviewed is hard work so try to make it as pleasant as possible. Eat lunch with them. Give them a t shirt!

Have them hack on a 1-3 hour onsite project working on your problems. They are allowed to ask questions, of course. You only want to do this if they’ve done well on the interviews and you expect them to do well.

After the on site interview you should be ready to make a hire / no hire decision.

How do you decide if you should make an offer? The obvious aside, there are a few things you should be aware of.

Setting the bar too high

People often preach the virtues of having high hiring standards. There's lots of theory out there about this but it's true in practice, too. Companies die regularly because they try to "staff up". That's undeniable.

But there is an opposite side of the coin that people don't talk about. 95% of hiring managers need to be told to "have high standards" and they'll iterate into the sweet spot from there.

But 5% of people need to be told to "have lower standards" and iterate into the sweet spot from the too-high direction.

We were one of those five-percenters. Xobni was originally VERY picky. There are a few reasons we loosened up.

The most important reason is that above a certain threshold of signal it's just impossible to predict how well someone will perform on the job.

There are so many random variables like your product, your team, your technology, your culture, their experience, etc that go into success or failure.

I've been surprised in both directions before. There have been people I had high hopes for who didn't work out. We've also hired people who have performed way above expectations.

So if you are too picky you eventually just start measuring noise.

Moreover, your interview time has diminishing returns. If you learn one unit of info during the first 60 minute phone interview, you'll learn only one more unit during the next full day of interviews. The next unit will come from a week of working together. And so on.

If you're interviewing a cofounder by all means spend two months working together before making a call. But otherwise your candidates will expect you to make a decision inside of eight hours of interview time.

Reminder: most people should probably have higher interview standards. Just be aware of pressures in the other direction. They do exist.

Hiring entrepreneurs versus hiring for patience

When we first took money from Vinod Khosla he told us "All of the first 20 employees at Sun went on to start their own companies or become CEOs of major companies. You want to hire those kinds of people."

I agree but disagree.

If you're a Sun or a Google or a Facebook then sure, hire these hyper ambitious folks. Or at least consider it.

But if you're not a rocketship, and most companies aren't, then be careful with hiring someone who's hyperambitious and thinking on short time scales.

The problem is that building a company takes a long time. It's a long haul. And you need the early employees to stick around for the long haul because they understand the broadest breadth of your company's culture, early decisions, code base, customer development, and so on.

So if your first hires want to start their own companies in the next two years you're not optimally set up for success.

Ambition is good in early hires, just not combined with antsy-ness. You want to invest in those people's companies, just not hire them for your own.

Your culture

So you really want ambition combined with patience. How about culture? How do you hire for that?

I'm not sure. I knew from the beginning how to spot great developers and have picked up bits and pieces about identifying other skill sets.

But I'm still working on hiring for culture fit. Here are a few examples I've been learning from.

Example one, just having fun.

Greg Thatcher is one of our many great developers. He runs a web site called gregthatcher dot com. He has several apps on there including a bank routing number lookup utility. Gregthatcher.com often comes up in office jokes.

Now check out gregthatcherdotcomfanclub.com, made by our very own Ryan Gerard. How fun is this stuff?!?!

I don't know the formal definition of company culture but even if this is completely unrelated I still wanted to share it because it's ninja awesome!

And that's part of why I come to work every day.

Example two, traditions. Every Friday we order the best pizza in San Francisco into the office for lunch. It's one of the things I look forward to every week. Yes, I'm that easy folks.

Example three, recognition. Beyond our universal recognition of gregthatcher.com as a jewel given to humanity, every Thursday at our company meeting we give out the Zombie award. It's given to an individual who has done something awesome every week. Everyone is encouraged to email Skyler, who organizes the award, throughout the week when someone does something outright baller. Then the winner is selected through an intricate set of informal procedures -- mostly banter and wit. The award comes with a large rubberish lego brick that the various offices can accrue and assemble.

I know, sounds a lot like the infamous Microsoft Ship-It award. It really is much cooler, though -- trust me.

All of this "culture" creates trust among your team, which is great for productivity, retention, new hiring, karma, fun, and your soul!

How do you build such a culture? Look for people who are passionate, positive, and creative. Then ask them to help you add spice. Empower them where appropriate, and wallah! That seems to be the formula Zappos has followed, anyway.

If you want a recipe for creating mistrust, hire people who are always cynical. That will weaken an organization like nothing else. They will drag down others around them, which gives them positive feedback so they make cynicism a stronger part of their identity, and the spiral downward continues. That won't end well.

Netscape's really-bad-attitude might be a counter example, but I don't think so. It seems to have been more of a way to blow off steam than to fester. Not sure.

Speaking of which, I'd love to hear war stories or lessons from folks in the comments on building culture. I'm very interested in learning more here!

Part 3

...is coming soon! Feel free to subscribe via RSS so you don't miss it!

I'm going to talk about how to woo and close candidates you want to hire, the basics of figuring out compensation, roles, titles, and a couple of other meta details that I hope will be helpful!

10/26/2009

How to Find and Hire Amazing People, Part 1

Nothing matters more to your success than the skill and determination of your team. So, how do you find and recruit these folks who will make you incredibly successful?

First, what experience have I had hiring and building teams? Moderate. I've directly hired about twenty folks and have been involved with but not responsible for hiring about ten or fifteen others. I've found that you can hire well without experience but you'll be about 3x more efficient with experience and you'll call fewer false negatives. So I hope to describe some of the patterns I've picked up over the past four years.

Prime Directive

Your first job is not to hire bad people. Prefer false negatives to false positives. I'll come back to this later but it's an important foundation to the whole discussion.

Finding Talent

The easiest way to find talent is through your networks. I knew this going in but didn't know how to reduce it to practice.

As an MIT engineer I don't have natural networking skills. Some think this means it's hard for me to mingle at social events, which you get used to, but the more damning result is that you don't have the mental gestures of thinking in terms of your "network." Network itself feels like a dirty word.

But network just means your friends. So when you need to "find talent through your networks," that just means reaching out to people you did school projects with at MIT. It's not really that bad.

Paul Graham originally told me to "shake my friend tree" when looking for a cofounder, which is certainly friendlier terms than "use your network." So shake your friend tree.

I can only think of two real group projects I did at MIT, and both of my teammates from one of those projects are or have worked at Xobni before. How awesome is that!?

If you don't have a network, hire folks who do. I hired three or four pros from my immediate friend tree into Xobni, but we've hired 15 to 20 folks from networks in general.

Example: we hired Bryan Kennedy after meeting him through Y Combinator. Bryan is our founding web engineer and is the mastermind behind all of our web properties.

Bryan referred in one of our next employees, Ryan Gerard, through a friend of a friend. Matt, my cofounder, and I were having lunch one Sunday and we got a call from Ryan. We gave him the Xobni pitch over the phone and the rest is history.

This is how it happens.

It seems remarkably inefficient that most hiring happens through friend trees and not through more merit based approaches. We'll probably have better approaches in 30 years but for now this is what we're stuck with.

Recruiters and job boards

Recruiters and job boards don't work well either. They are supposed to fix this problem, and despite the gobs of money people are willing to pay they are still unable to deliver in general.

First, bad people (heretofore "unqualified workers") stay on the market longer and cycle their resume around 500x more often than a great employee in the lifetime of their career.

I kid you not. How many times do you think the top 1% of developers send their resume to companies? Maybe 15 times during college, they get an initial job, and get to pick where they work from there out based on reputation. So maybe 30 resume-sends over their lifetime.

Whereas unqualified workers will be on the job market say 15 times, each time spamming their Java resume (sorry -- couldn't help myself) out to 100 places each time before getting hired.

Someone check me on the math, but it sure feels like a 500x difference.

Recruiters won't save you, either. They get paid when you hire someone. They do not PAY YOU for consuming your time.

So a classic first time entrepreneur mistake, which I made, goes like this. You raise a bunch of money, and the recruiters come calling. Paul Graham didn't talk about recruiters, so you don't know what to do. (!) You listen to their tale, and they tell you they don't get paid unless you hire one of their people. So you can't lose. You sign a simple contract and go to sleep.

Then you wake up the next morning with seven resumes in your inbox. You'll get another seven the next day. And they'll keep coming until you interview someone. They will wear you down.

What happens next determines whether your company will fail or live on to fight other battles.

If you are looking to "staff up" you will look at the number of resumes you've read, and you'll say to yourself:

"Self, you've looked at 200 resumes and interviewed ten of them. You've worked hard. You certainly deserve to pull the trigger and soak in some celebration and accomplishment."

And repeat until you've hired an engineering team full of people playing the part and burning your cash.

Seriously, I've seen a company do this. Needless to say they sank without a trace.

Instead, if you maintain your hiring standards you will wisen up and dump the recruiters. That's what happened to me but I hope you can one-up me and avoid spending time on recruiters overall.

Now this isn't true of all recruiters. Your mileage may vary. What usually happens is the good recruiters move up and do executive recruiting, so most staff recruiters are crappy. I don't have enough direct experience with executive recruiters to speak about them.

There are good staff recruiters too, but they are very rare. At the very least ask for some clause to be changed in their contract and see if they have the organizational power to do it. If not you're dealing with a chop shop / call center. It's also a bad sign if they anonymize the resumes they send you. Most do.

Back to job boards: I have found that the Joel On Software job board doesn't have the spam problem. A typical Xobni post there nets us two good resumes, one bad resume, and that's it. Not cost effective but at least not spammy.

Craigslist can also work for hiring some roles, particularly where skill is easy to measure and there isn't much variation among choices, i.e. not developers, not product people, and not business people.

When you do use Craigslist run a bulk recruiting process. Post once on craigslist, ask everyone who emails you to answer a few questions on a form, and invite those that respond and read well to your office at thirty minute intervals across one day. Have all of the decision makers available and at the office on that day.

Finally, use jobs@yourcompany.com on your home page but use different email addresses for any job board postings you do. Blind emails to jobs@ from your home page will tend to be high quality, so you'll want them in separate buckets.

Finding folks using your network

Building a network and using it to find the best people is a long term play. CEOs who knew lots of people at Yahoo got a windfall when Yahoo started faltering, for example. These events are rare and hard to predict.

You also can't predict when you'll get a phone call from a friend of a friend who isn't happy at his current gig.

The best you can do is try to encourage these random events. Meet lots of people, spread the word about what you're doing, and be so awesome and excited that people remember you and your enterprise.

This is a great strategy overall, and will result in more frequent coincidences that result in positive results.

For more on this topic, go read Never Eat Alone and this blog post.

Loose in the socket

I originally assumed that people's happiness at a job falls along a spectrum, maybe in a bell shaped curve. Not so. It seems to be binary: either someone is happy, or they're not and open to a switcheroo. I call the latter group loose in the socket.

Also, the really good and intrinsically motivated folks won't come work for you because you offer a higher salary.

At the end of the day, it's disproportionally harder to recruit someone who is happy than someone who is loose in the socket.

This is not to say you shouldn't talk to good people you know who are happy in their current job. Just realize they are long term plays as mentioned above. If someone is happy where they are, your best shot is to (a) have a really compelling product and team, (b) tell them you have an open door for them and would love to tell them more about what you're doing at any time in the future, and (c) hire their friends or people they know are good. These long term bets can work over the long term, and are bets worth placing, but you can't count on them in your pipeline.

Part 2

Part 2 and beyond will be coming soon, and will touch on topics like:

• How to interview and judge candidates (very important),
• How to close,
• How to resource and time your hiring efforts,
• How to recruit for culture, and
• How to negotiate and set expectations,

...among other things! Check out the RSS feed to stay in touch!

10/21/2009

Early Startups Needs Generalists

I was at MIT for a few days last week for the MIT Startup Bootcamp. One thing I love about MIT entrepreneurs is how scrappy and motivated they are. Compared to the iconic facebook chasing social media west coast entrepreneurs, these guys are Rocky Balboas. They stay in the game when others would have given up, and they tackle substantive problems.

MIT graduates still come out with several learned habits that will work against them though.

One such habit is to obsess over everything. At MIT there's a very finite set of material to study at any time in any given class. The hard core students among us make sure that the curve is attrociously high. You might spend 20 hours studying a small set of material.

So the onus was on expertness. Of everything.

This doesn't work in an early stage startup where everyone has to be a generalist and work on seven different projects per day.

You don't want to be an expert in your series A docs. If you are negotiating the finer points in a legal negotiation, you are losing.

If you are worrying about your computers costing $800 versus $1200, you are losing.

I know the MIT obsessive in you wants to optimize everything, but just don't.

Instead, hit REALLY high notes on the important things.

HubSpot is a startup that's all about bringing modern Internet marketing to small and medium businesses. It's a huge opportunity and a huge deal. Their biggest challenge is educating the 100 million small business owners about this new world.

For HubSpot, educating their market is incredibly important to their success.

So they put on a conference. Their founders wrote a book. They've released several free and easy tools like TwitterGrader and WebsiteGrader. They take this stuff seriously, and they should. Their MIT'ness is shining through in the best of ways.

So, for folks like me, half the battle is identifying where to hit the high notes and where to swing for the B team. And when it comes time for the B team, swallow your pride and show 'em how impressively unimpressive you can be.

10/15/2009

13 Ways Acting Classes Improved My Public Speaking Skillz

Now, I am no public speaker. In 2007 I got a chance to say some words at Startup School along with several other Y Combinator founders. My voice was cracking and I spoke at ten thousand words a second I was so nervous.

Which is why I was excited to give a talk to eight hundred people at this last weekend’s MIT Startup Bootcamp. This was a chance to redeem myself. I’ve been taking acting classes for the past six months to work on my communication and presence skills, so this was a perfect mid-term exam.

I put about 30 hours into the talk. I wanted to test how far my skills could take me so that meant putting in an excess of time to make sure I took the best possible shot. I wrote a full twelve page script that I memorized. I rehearsed with my acting teacher for three or four hours. I spent a hours on a couple different versions of my slides.

I’m calling this one a success! It was the least nervous I have ever been doing public speaking. Part of that might have been that I only got three hours of sleep the night before and just had no energy to be nervous; I’m not sure. I had gone to sleep at 1am, woke up at 4am, and couldn’t go back to sleep until it was time to get up!

But, without further adieu, here are the thirteen things I learned that went into my public speaking improvements, in no particular order:

What do you think? Any other good tips for MIT personality public speaking?

09/13/2009

MIT Students Send Cameras Into Stratosphere, Very Cool!

I got an email yesterday about some MIT students from my fraternity creating a balloon for $150 with no custom electronics that took pictures from 80,000 ft in the air.

07/08/2009

Wireless Data Everywhere, M2M Communication, and the Radio Tagging Problem

In defiance of California state law, when I hear a new jam on my car radio I pull out my IPhone, launch Shazam, and use it to "tag" the song. Then I scroll down, press "Share This Tag," type in my email address, and return my eyes to the road. Any number of days/weeks later, I download the mp3 and have it added to my collection.

06/10/2009

Some Thoughts: the Online Backpacking Travel Industry

A friend recently wrote and asked for opinions about a web site idea he had. The idea was "yelp meets international backpackers" - reviews for restaurants, places to stay, etc. The tripadvisor for post college travelers.

Howdy,

Rock on dude! I've always been curious about the backpacker traveling industry so this is a cool way to find out more about it.

I looked on compete.com and compared the traffic of hostelworld and couchsurfing. They each get about 100,000 unique visitors per month. Hostelworld apparently has 20 employees based on their home page. I'm not sure if they are all local or international / what their average wages are, but I'd venture to guess that their burn rate is about 2M per year, and I'd guess that their revenue is just north of that...maybe 2.5M or 3M per year, optimistically. If they have 1M users per year (somewhat optimistic based on their monthly traffic, but (a) they probably have lots of turnover in their user base, and (b) multiple people using the site from the same hostel probably only counts as one 'visitor') that's about 2.5 dollars in revenue per user per year (ARPU), which sounds about right given how close they are to the transaction.

Couchsurfing died a couple of years ago in a major data loss and only started back 'recently' so their growth is a little bit misleading. Anyway couchsurfing isn't in the middle of a transaction so they almost certainly have a harder time making money. They ask for donations essentially.

Much better to be airbnb than couchsurfing, if you're in it to build a big business. Airbnb is couchsurfing but with financial transactions. They got started with $15k from the same investors who gave Xobni our first money, so I don't know them but they are very close in my social network.

Anyway, it's just helpful to understand the comparables so you know what you're getting into!

Yelp on the other hand has 25M unique visitors per month, and growing! The yelp CEO mentions international growth here: http://news.cnet.com/8301-13577_3-9884023-36.html. More likely, though, they are more focused on generating revenue from those 25M users much more than growing internationally.

Anyway, I would focus on the following questions:

1) how will you acquire new users?
2) if successful, how will you make money?

The risks there seem large, but they always do at the beginning of any successful venture. :)

06/02/2009

Quick Software Company Idea: Remote Real Time Help