Jul 14

Gmail’s March to World Domination: The Platform Strategy

Gmail plus Chrome Gmail is an app eco-system that has snuck up on me. I now can’t use Gmail without:

  • Assistant.io – An amazing little meeting scheduling app.
  • Boomerang – An app that lets me schedule emails to be sent later or returned to my inbox as a reminder.
  • Rapportive – Lets me see info on the people I’m emailling.
  • Relate IQ – Lets me update our CRM from within the email.

And thats not even getting to the useful Chrome extensions:

  • Honey Badger – A great tool to learn about the traffic and funding of a website.
  • Evernote – I use the web clipping to make articles easier to read even if I don’t plan to save them.
  • Ghostery – Tells me what cookies are being used to track me on a given page.
  • LastPass – Secure passwords, nuff said.
  • BufferApp – Allows me to schedule my social media posts.

Gmail is slowly building a platform that will be very difficult to migrate away from. I’m sure others said the same about Outlook in the past, but cloud platforms seem more powerful than their desktop ancestors since they so easily build upon the services provided by others.

Jul 14

Too Cheap To Count

hash marks

Fred Wilson has a great post about what technology was rare enough for one generation to count but ubiquitous for the next
• Electric Motors 
• Chips 
• Networked Devices 

He also talks about how the networked devices of the future will be dumb. 

So my bet is that most “things” will be dumb and the smarts will be in the phone or in the cloud.

But I respectfully disagree. In my post about computing paradigms swinging like a pendulum I stated that Moore’s Law means that dumb devices become smart over time.

via Intel presentation

via Intel presentation


While I expect Fred to be correct in the short term, dumb devices connected to smart hubs, over the long run I expect each of those devices to be imbued with more intelligence than Siri has today

I believe that while you and I know the number of “smart” devices we own, the next generation won’t be able to keep count. 

Jul 14

History is All Lies: The Pendulum of Computing Paradigms

As with many of these posts, I learned that history is often apocryphal when I go to link quotes. I started by trying to find the origin of this famous quote:

I think there is a world market for maybe five computers” – Thomas J. Watson

This is probably a misquote, but it is a great start to the pendulum that is the centralized  vs. decentralized computing model. Let’s walk through the 3 major computing paradigms: Big Iron, Enterprise Computing, and The Web, after which we will look at where we stand today with Mobile Computing.

Big Iron: In the beginning we had centralized computing. Giant machines with an army of attendants that fed in punch cards. Then we attached dumb terminals, teletypes, to it and called it Time Sharing. This was centralized.

Enterprise: Slowly the dumb terminals started getting things like storage. Moore’s law took effect. We got computers on the desktop and the model was Client Server. While there were some shared resources, computers generally had all their software and data locally. This was decentralized.

The Web: When the web was born we had light weight browsers that connected to powerful servers once more. Once again all the resources resided in the center, and the browsers were essentially dumb. With the innovation of AJAX the browser started doing more work for the user, but it was still essentially a dumb client. Centralized.

Mobile Computing: At first phones were dumb; but soon  Moore’s Law kicked in and the devices became powerful enough to do real work. We even renamed them Smart Phones. This gave rise to application paradigm. Once again, the phone ran software and much of the data needed was stored locally in case the network wasn’t accessible. Decentralized.

As long as Moore’s Law holds, which I believe it will, this pendulum will keep swinging back and forth as the dumb devices begin to have enough computing power that it becomes worth the effort to harness them. I believe this has some interesting implications about the different ecosystems that Apple and Google have created, the former being more device oriented and the later focusing more heavily on the cloud.

May 14

Startup Memes: Github for X

<I actually wrote this several weeks ago, but then life got in the way. I think it is still relevant.>

This is the first in a (hopefully) recurring theme of posts discussing patterns I’m seeing in startup companies.

For the past few months I’ve seen a number of companies that can be described as Github for X. Basically an open storage platform where people can fork off projects. While I’ve seen a bunch of others, some good examples are:

Partly this is probably due to Githubs big fundraise. But another aspect is probably that coders are starting to branch out to other domains and they are trying to bring their tools with them.

Back in the dark ages we used RCS & CVS to track source changes. They were hard to use in teams and didn’t scale well to the Internet. Today we have incredibly powerful tools with integrated sharing platforms.

For those who’ve used these tools with a distributed team you know how powerful they can be. It is only natural that people will want to bring them to other domains.

Apr 14

Coder Whisperer or How to Energize Your Dev Team


Given my  background, an odd thing happens when I am the only technical person in a business discussion; I am often turned to as the code whisperer. Non-technical founders and board members will ask me how coders will react to something or what would get them excited.

The problem: while many people are extrinsically motivated (from the outside) by things like money or responsibility to the team, most hackers are motivated by interesting problems. Smart people understand this; what befuddles them is what makes a problem interesting to a hacker.

Things that might seem trivial to an end user, a game of boggle or sorting a list of names for instance, can house some very interesting puzzles for a programmer. To make matters more difficult, not all coders find the same types of problems interesting. A seemingly mundane problem may give a hacker the chance to explore a path they are curious about. haystacks Think of Monet’s Haystacks, it is highly unlikely that the painter had an innate interest in the making of hay, instead he used the haystacks as a way to explore the nature of light and how colors changed throughout the day. Programmers can do the same sort of exploration within a problem. For instance, they may use a project to learn a new programming language or explore a new development framework. Some common interest areas:

  • Algorithms
  • Scaling to large numbers of users
  • Using new languages or frameworks
  • Data Science
  • Optimizing an application to run fast

This isn’t to say that you can cleverly wrap up a problem and fool someone into working on it, instead this gives you the best chance of finding problems that your programmers will want to work on. Like most people, hackers want to work with great people. So the best way to get a great team of hackers is to start with one alpha hacker. But knowing how programmers think will give you a huge leg up in the hiring department and let you find projects that will excite the best and brightest.

Sep 13

Should CEOs Code?

coding_punch cardsI recently sat down with Peter Bell, an excellent tech trainer, to discuss an array of things. I had in my mind a prototype for this blog post that Peter promptly sunk.

My original idea was to write about outsourcing development and why I hate it ( I still do), and the stance I was intending on taking was “Hey ‘business-guy’ learn to code!”

So how did Peter change my attitude? He pointed out that much of learning to program is syntax, and neither of us thought this was important for a CEO to know. What I wanted was for the business guy to be able to speak the same language as the devs, and to have some intuition about which problems are hard and which are easy. In my CS program at Purdue this was mostly covered in the 2nd or 3rd CS course, after we had taken an intro to programming course.

Peter is actually working on a class to teach these skills to business folks, and I respect that a lot. He actually has a couple of questions he suggests people ask when trying to find a technical founder.

What revision control system do you use and why did you choose it?

One doesn’t need to know the answer but instead listen to the thought process of the engineer. A good engineer will have tried a few solutions and have chosen the system for a couple of salient reasons. A bad or novice dev will have chosen based on ‘everyone else uses it’  or ‘ it’s what I was taught’.

Peter has a number of other such questions but the principle remains the same: ask a question about technology and listen to the process the developer used to come to the answer. If that process was rigorous and thoughtful then you’re probably talking to a pro.

While I still think outsourcing development is a terrible idea, I am now less convinced that an intro to programming class is a complete answer. One really needs to understand CS design rather than a programming language. Of course, if you’re young you should still learn to code before Skynet takes over.

Jul 13

The Age of the Cypherpunks has Finally Come

Let’s start with an admission, I’m older than I look. Back when I was in high school there was an online group called the Cypherpunks. From Wikipedia:


Cypherpunks originated as an informal online group of people interested in privacy and cryptography who originally communicated through the cypherpunks mailing list, although there were also cypherpunk meetings and parties in real life.


The cypherpunks mailing list had extensive discussions of the public policy issues related to cryptography and on the politics and philosophy of concepts such as anonymity, pseudonyms, reputation, and privacy.


Apparently there is still an active node, but usage really peaked around 1997 or 1998; and I followed those discussions religiously. People were talking about really cool ideas around digital commerce and anything seemed possible.

Well, it looks like at least a few of those ideas have finally become a reality: Digital Cash and the Ransom Publishing Model.

Digital cash is pretty obvious, and it looks like Bitcoin is getting enough traction that we’ll at least get to see an interesting experiment play itself out. I’ve seen the potential in truly anonymous digital cash since I was in high school and I’m psyched to see people using Bitcoins.

Back in the day I worked on the Mnet Project, a distributed file share that used crypto and had come out of a micropayments file share called MojoNation.

Mojo was a digital cash currency that aimed to provide attack resistance and load balancing in a fully distributed and incentive-compatible way (see Agoric computing).

Long story short, I really wanted Mnet to become the basis for an online digital currency, but it never gained traction. But now we have Bitcoins, and I’m pretty excited to see what awesome ideas people come up with using true digital cash.

The second Cypherpunk’ish idea to gain traction is the Ransom Publishing Model, also known as the Street Performer Protocol. John Kelsey and Bruce Schneir released  a paper on this back in 1998, but people had been talking about it for a long time before that.

ABSTRACT: We introduce the Street Performer Protocol, an electronic-commerce mechanism to facilitate the private financing of public works. Using this protocol, people would place donations in escrow, to be released to an author in the event that the promised work is put in the public domain. This protocol has the potential to fund alternative or “marginal” works.


Sound familiar? Yeah, it’s Kickstarter.

I’m going to go back and revisit some of the other Cypherpunk ideas that had me really excited back then; I believe the world is finally ready for them.

If you have a Cypherpunk’ish idea whose time has come let me know, I’d love to chat about it.

Jan 11

FB: Take the Money and Run

Tyler Shields (@txs), an old friend and colleague,  asked me the following:

I know you have some education and experience in the business world so I was curious what your thoughts were on something. I was debating with some folks about the Facebook valuation and what the results of this Goldman’s Sachs back channel IPO is going to be. If you had the chance would you throw money into the pile or would you wait and hold? Also do you think they are overvalued or fairly valued at approx 50B$?

So I first took this to mean would he invest if he had access to GS, then I figured maybe he meant would you dump if you were an FB employee. So I have both answers.

Let’s start with value. (Since that will inform whether you want to put money in.)

Some context-

  • Google is worth 200B.
  • AOL is worth 2.5B
  • Yahoo is worth 21B

Now everyone who has bet against high FB valuations in the past has been wrong. And I can see that FB might be worth twice what Yahoo is, and they have just surpassed Google in traffic. So Maybe 50B is a fair price. BUT, part of the Goldman valuation has to do with the fact that they can make fees by selling parts of that 1.5B fund to their clients. And they will likely get to do the eventual IPO, which will earn them more fees. So Goldman has a revenue stream that most of us can’t count on.

Possibly the most telling data point would be Digital Sky Technologies, the Russian firm. They are already an investor, and they have a pretty big stake. If anyone has a decent idea of what’s really going on it’s them. If they put in more money, then you’d hope they did their due diligence. So if I’m arm chair quarterbacking I say it is probably a fair valuation.

On the other hand, if I really had 10 million and GS came to me I’d say no. The reason is simple, you don’t get to see the books (the way GS and DST did). If you can’t do your own due diligence then you’re a sucker; pure and simple. Sure, you might make a lot of money, but you can also make a lot by betting on red in Vegas. Doesn’t make you smart.

Now, if you’re asking what I’d do if I’m an FB employee:

This is actually an easier answer.


Though it always depends on your personal situation, most people would tell you to take some money off the table. What happens if the company explodes? You not only lose your job, but also your nest egg. That’s not smart. The common answer is to not hold your companies stock (everyone cites Ennron here). But this is FB, the hottest investment in the world. So maybe you limit yourself to 10% of your nest egg… If you’ve got 10M in stock, I’d dump as much as they let me. (I think it is 10-15%) Why not take some of that risk away.

Also, you have to expect that you have a ton of unvested options. (Maybe not, but my guess is they use ’em to keep you there.) So even if you sold your vested portion, you still have plenty of exposure.

There you have it, my un-asked-for advice to current FaceBook employees.

Oct 10

Everything Old is New Again: WildFire is the Web 2.0 Version of FlashBase

The first startup I worked for, about a year after graduating college, was a small company named FlashBase located in NYC. I believe I was the 8th employee and the 3rd full time programmer, not counting the two founders who wore many hats.

FlashBase started life a web front end for database backends, which was a technology in search of a solution. Basically we allowed users to graphically create webforms; common uses were surveys, help boxes, and sweepstakes. During the attempt to raise a 2nd round of funding we made a pivot, and switched from a general tool to a specific solution: Sweepstakes.

DCLK Sweepstakes
Thanks to the wayback machine we can see the homepage a month after the aquisition. (The pre brand change site doesn't have images and hence looks pretty ugly.)

It turns out that getting opt-in mailings lists was important enough for people to actually pay for our product. And thus a business was born. We made a turn key solution for running sweepstakes. Which is why WildFire sounds so familair to me. WildFire is a platform for running sweepstakes and doing analytics around them. Essentially, it is the Web 2.0 version of Flashbase; better analytics and tools for virality I'm sure, but essentially the same product.


While web design has come a long way, the value proposition is the same: WildFire creates turnkey solutions for sweepstakes, user-generated video, photo, & essay-based contests, and coupon give-aways.

In the end, just as the dot com bust happened, our founders sold the company to DoubleClick for around 20M. Given the peak for the NASDAQ was in March selling in May was quite lucky, the slide had only just begun and people still thought that a recovery might happen. For the VC, Dawntreader, this was a pretty decent outcome since so little had been invested and the flip was quick.

So what does this mean for Wildfire, who has taken 4M? I suspect it is going to be a tough road to hoe. A 20M sale aint what it used to be, and buyout valuations are lower now than in 2000. I'm sure there is an arguement that Social changes everything, but having been there in the past I don't expect WildFire to be a "Big Company". For a month long campaign at the premium level ($250 + $5 a day) we are looking at $400. Let's say they can sell the company for 3x Revenue. So to reach a 20M sale they need 6.6M in revenue. To get that at $400 a pop you need to run 16,500 campaigns a year. Let me say, that's a lot of sweepstakes.


Jul 10

Lean Startup: Early Testing Using Ads and Splash Pages

First, a bit of background. While in school I had this idea about a tool to help with networking, something to replace the excel spreadsheet I was using to track contacts. Recently the urge to scratch that itch became too great and so after years without writing any code I picked up a book and began to learn ruby. I spent a few weeks over Christmas coming up with a bare bones prototype, and I was feeling pretty good.

Jason Freedman, the CEO of FlightCaster, and I were having coffee and I told him what I was up to. His immediate reaction was “have you read 4 Steps to the Ephiphany?” Since I hadn’t read it at the time Jason went on to quickly explain that I should try to iterate my idea as much as possible, and that I could begin without any code. His suggestion was to just run some adwords and see what the response was.

Since one can’t run adwords without at least a landing page, I went and grabbed an open source website template and put it up on a server I had. I added in the Google Analytics code so I could track site traffic, and I was ready to try some Ads. My goal was to cheaply drive some traffic to the splash page to test my value prop.

The first ads I created didn’t do too well.

Engage Your Network
Make professional networking easy
FREE Tools for keeping in touch!

Though my conversion cost was low, around $.050 the adwords I could afford didn’t get much traffic.


Keyword Status  Max. CPC Clicks Impr. CTR  Avg. CPC  Cost

Some were really quite good with 5% CTR but only a handful of impressions. With only 6 clicks over a few days this was going to take too long to get any usable data.

After doing some Google research I found a great post by Ash Maurya talking about landing pages and advertising. I decided to give FaceBook a try. I needed a graphic to use with my ad so I went to istockphoto and spent a couple dollars on a small pic that seemed to fit. I tried several ads, this was the best CTR, though the bounce rate on all of them was 100%.

Keep your network engaged
with FREE tools that tell you
when to reach out to
business partners & friends
that you’re losing touch with.


Campaign Status Budget/day Clicks Impressions CTR (%) Avg. CPC Spent
25 165,989 0.015% $0.54 $13.45

Now I was getting 25 clicks in a couple of days, and this was limited by my budget and not the impressions for the adword. But the 100% bounce rate meant that either my value prop wasn’t there, or that FB ads weren’t the right way to reach my audience.

Steve Blank says that you’re trying to sell the product you’ve got, rather than modify the product to fit the sale, so I decided to try another group. I went over to LinkedIn since this was pretty much the demographic I was after, people who are interested in networking with an eye towards business.

Here I started to see something a bit more promising.

Date Impr. Shown Ad Clicks Profile Clicks Total Clicks Ad CTR Total CTR Avg. CPC Avg. CPM Total Spent
2010-02-16 16430 9 0 9 0.05% 0.05% $2.00 $1.10 $18.00

So the CTR still isn’t great, but now I wasn’t getting 100% bounce rate, in fact it was 55%.

Now my page was really very simple. It really only consisted of a splash page and a learn more page. I recently added a beta sign-up page, using Google docs, and even got a couple of people to fill it out. But doing that begs a few questions. What should I ask them, and how should that call to action be done. For instance that should probably be moved to the splash page.

So what has this little experiment taught me? Glad you asked:

  • Adwords is difficult for a product that isn’t in an established category since people don’t know to search for the product. (I’d have to do more thinking about carving a niche.)
  • Finding a place where my demographic hangs out worked well. (Not positive this is generalizable, but it might be)

The next steps were pretty clear. I had to figure out a call to action splash page and use it to find “beta users” willing to fill out a short survey. Some A/B testing might be in order as well since users weren’t immediately bouncing.

Really was just a cheap way for me to see what tools might work going forward.