(this is page 11 of 121)





Latest

Ken Olsen, founder of Digital Equipment Corporation (DEC), and by extension the east coast technology corridor, died this week. Though he was a very low key guy, the legacy of the company that he built still looms large around here, with both the engineering culture he created and scads of alums having gone on to live well beyond DEC's being swallowed by Compaq over a decade ago.

Of all the commemorative stuff written about Olsen this week, it was Andy Kessler's piece at GigaOm that most caught my attention. In it Andy (who often writes really thoughtful stuff) ascribes the failure of DEC to its inability to "horizontalize" in the face of the PC where Intel, Microsoft and the banana handlers (HP, Compaq, Dell, et al) focused on providing just one layer of value instead of DEC's attempt at an integrated solution.

It would certainly be easier if it was truly axiomatic that this type of vertical integration is certain death for companies in technology. Except of course that it is not as in the case of the smartphone shows today. Or what the storage vendors have shown over the last 10 years. Or the current "cloud" providers on both the compute cycle and storage services.

The best model for understanding what really matters is presented by Clayton Christensen, of Innovator's Dilemma fame, in a book called Seeing What's Next (which in my opinion is the best of the lot because he puts his framework to practice). In it he fleshes out the argument that vertical integration actually makes sense depending on how effectively the technology can address the market need. In cases where some fundamental constraint (computing power, storage, cost) exists to deliver what it takes to satisfy the customer, it is often the integrated players that win. And when the technology (often fueled by the exponent in Moore's law) gets to the point where it overshoots the solution, then it is ok for layers to be created where each one can begin to move at a different rate, and incidentally when the most opportunities are created for lots of capital efficient startups to be born.

But integration is not bad per se, and more importantly, it is not always a quick transition from one state to the other.

In fact, this transition was not what killed DEC. DEC was vertically integrated in the minicomputer space because it made sense to be— even long after the PDP era, the tight coupling between VAX and VMS made it possible for the company to use relatively underpowered hardware to deliver solutions into the market that were decades ahead of the rest of the world. What killed DEC was much simpler: they just missed the PC altogether (the DEC Rainbow was a half-assed attempt at getting in the PC business).

So what is the lesson to take here? First that over simplifications are often wrong and in the era of blogs an retweets they can often hide simplistic thinking in easily digestible repetition.

More importantly though, to all the folks waiting for the "horizontalization" that got DEC to come and kill today's integrated king, or who perhaps are pointing their product development roadmaps towards the coming "layer" winners (read: Android), you might be waiting a long time. Or at least until the BrainPal comes to kill the smartphone.

Comments Permalink

Computation put to good use

Having recently spent sometime behind the big tech company iron curtain, I feel fairly inoculated against the rah-rah style corporate video that tries to render the profit-making motive in a romantic "change-the-world" form. However I keep coming back to IBM's 100 year anniversary film by Errol Morris:

It's a terrific reminder that computation has been put to use for much grander tasks than self-actualization and better shopping. Not that there is anything wrong with those things (I've spent a fair bit of my career mining those two), but it's hard to compare them to say, putting in the additional return-to-Earth calculations into the Apollo 13 computer that saved the astronauts.

Comments Permalink

Chris Zacharias, a YouTube front-end developer, had a very thoughtful post around the challenges of hiring front-end engineers. Despite having hired a lot of engineers during my career (as engineering manager, VP of engineering, founder, & CEO) and having honed a process for figuring out who was likely to work, the front-end (HTML, CSS, & Javascript) role has become the biggest challenge of the lot— especially after the 2004 "AJAX revolution" when the role became quite a bit meatier.

And it wasn't until Zacharias's piece that I realized that it's because I'd always used the wrong resume screening & interview approach. Because I've recently fielded a load of questions from entrepreneurs who also find this a hard role to recruit and hire for, here is a list of the "don'ts" that I had to learn the hard way:

Don't braintease them

When you are hiring an engineer who is working in your app tier or further back in the stack, brainteasers are a good shorthand way to test for good hygiene. Logic, efficiency, indirection, recursion— these are all important traits that can be evaluated through a variety of classic logic problems (my favorite, the Pirate Game, tests all of them at once). Plus, you also get to see the candidate think in real time and under pressure, both good proxies for horsepower and what startup life is like late at night.

The problem with brainteasers though is that they're algorithmic porn and in the best of cases they will fail to test for the right high level traits: tenacity, debugging ability, and the willingness to power through seemingly impossible implementation tasks. And at worse they will make you look like a clueless snob who thinks that Zuck talking about speed chess on 60 Minutes somehow relates in any way to writing the world's most scalable PHP hack.

Don't make them whiteboard sessions

Another classic mistake— the old "how about we think through how to implement X" where X can be a persistence layer, a protocol, a state machine, etc. You then watch the candidate sweating through pseudocode to check for design skills. Again this will work for people from the app tier on back, but taking any front-end person through this kind of thing will only result in generic discussions about the CSS box model and the pain that IE 6-X is. In other words, the crap they read on Zeldman's blog en route to the interview.

At Tabblo we quickly hit the limits of this style of test and realized that a) the time constraint was stupid and b) what really mattered was making sure the candidate had the dogged attention to detail required to make something work across a variety of different browsers. So we settled on a "take-home test" which meant giving the candidate an assignment after the first interview with about an evening to complete (in our case it was an implementation of a dialog box which fades up on a standard page without the use of supporting libraries like jQuery). We named the test the Hausse test after a very polished candidate named John Hausse* who came in with great references spewing all the right jargon about standards-driven web development and then completely failed (for years afterwards, we asked each other if candidate X had "Haussed" his takehome).

Hiring correctly is one of the hardest things a startup can do and the engineering team is the engine room. While the mobile revolution and the resulting return to native apps may make the new front-end hiring process simpler in the fullness of time, there are still loads of web work to do. If you are in the process of that, skip the age-old stuff mentioned above and do yourself a favor by reading the above mentioned Zacharias post.

* The name has been changed to protect the innocent.

Comments Permalink

What was great about open source in 2000 was that it was a small club. Sure, Cobalt has sold to Sun for a lot of money, and sure Red Hat and VA Linux were on their way to IPO boom-and-bust. But it was still the case that if you were reading Slashdot to get your Hauppauge TV encoder card to do the Tivo on your Linux box, or learning the new-fangled ipchains utility to get your Gnutella client to work, you were definitely in the land of the Morlocks. Just a bunch of geeks talking to other geeks about doing things that most normals couldn't even dream of— Pringles can antennas to boost Wifi signals and sniping cron jobs that always let you win at eBay— "magic" to the rest of the world.

This has all changed— in part because of social networks, and in part because of the smartphone. As Patton Oswalt writes in the latest issue of Wired, geek culture is now mainstream: it is "cool" for Zuck to go on 60 Minutes and talk about speed chess while your accountant next door neighbor tells you about the latest iPhone app that tracks fuel consumption. While the geeks were busy thinking they were inventing the world, the world caught up. And sadly, we geeks now find ourselves scooped by the guy who spends every night surfing the AppStore for the greatest and latest or the guy who is watching his Twitter feed for the demos from the latest Meetups.

What to do about this?

Well outside of the fact that it's important to realize that "geek culture" is now akin to talking about "Gap culture," we might go back to first principles and think a bit about what we were all after in the first place:

1. a network that connected anyone to anyone independent of existing structures of control
2. a way to consume the content (or anything) we wanted when we wanted
3. a way to make computers augment our own intelligence


On items #1-2, I'd say we ought to just give up the ghost— it's happened and it is televised, mainstream, and on Facebook. #3? Not so much.

So maybe the way to cut this problem is to let geek culture die on things like "on demand" and "connecting with people" and focus instead on the fact that we've not yet cracked the problem of computers truly augmenting our intellect. Not in the way that Doug Englebart meant back in 1968.

When I think abut the subcultures that might be most relevant to this, I think of artists using computers to create not better art, but entirely new art forms altogether (like personal fabrication or locative art). I think of product managers who think about immersive experiences and bringing together online and offline in new ways (Virgin "touchpoints" for the rest of our lives). I also think of all of the programmers working hard to create ontologies in places where dumb indexes that are blindingly fast give us the false impression that services can divine what we want (semantic web junkies rejoice!). And perhaps most importantly, I think about the folks that have already taken a broadband-connected pocket computer for granted and are trying to puzzle out what a world mediated by this crowd-empowered outboard brain looks like.

Put another way, I think the next relevant subculture coming to us is from China and India, and other places where the PC never stuck and the smartphone landed even more definitively. I think it may look like "For the Win" or something quite different. What it doesn't look like though is geek chic— 60 Minutes speed chess and Harvard dropouts talking about "utilities."

The geek subculture didn't die, it just went mainstream. And now it is time to look elsewhere.

Comments Permalink

The last few days have been a blaze of reports about the future of Android and the sub $100 smartphone in 2011. I've been sucked into pieces from Fred, Fortune, Horace, and even some random dude (with good points to make) just as much as the next guy which is why it behooves me to say this now:


If you live in the world of startups (as an entrepreneur, rockstar hacker, seed investor, VC, or even cheerleader) none of this inside baseball stuff about who is going to take market share in the smartphone wars matters to you.

There, I've said it. If you are like me and don't actually follow sports, this whole market share thing is a great substitute. It's also a big deal for the big tech companies; for instance, at HP we were highly motivated to have Android commoditize smartphones (at least until they went and bought Palm). Google also suffers tremendously if search doesn't port well to this new platform. And clearly Apple (which has already sold Macs to every sucker willing to pay the premium) loves the idea of selling more CPUs to the folks in the cult (me included).

But if you are building the next big SaaS startup? The next social commerce killer app? The next Internet infrastructure bit? You should absolutely not give a shit at all.

In fact, let me tell you what the recipe for determining what your "smartphone strategy" should be:

1. Think hard about whether you can ship an HTML5 version that your customers can find, use, and be delighted by. If so, do that and skip the rest.
2. If the answer to #1 is no, build an iOS app.
3. After you get to a million downloads (yes, 10^6) think about writing a version for Android. But only after you GOTO #1 and re-evaluate that issue.

Before you stop reading, consider this: I have 3 Android phones. I worked on several Android projects at HP. I love Android, but I love it as a political statement only because Apple's controlling stance on everything creeps me out. If you want users with credit cards that spend money, mainstream reach, or a platform that actually might make an interesting case for abandoning open standards ("write once, run almost anywhere"), then iOS is it-- for everything else, just be happy the mobile web is now a legit platform to deploy to.

I meet with so many entrepreneurs who run startups and say something like "yeah, we love Android" and then go on to give me a demo on iOS. Why? Because this is where the users are. Not because it is more polished (it is), or because it has a higher quality filter in the AppStore's approval process (it does), but because when you are running a startup, you've got limited resources-- and navel gazing about whether the $85 smartphone is going to cause Android to "explode" doesn't help to get you closer to winning, nor does betting precious development resources on political causes (backing the more "open" guy).

But wait, doesn't the openness engender more startup opportunities? Try as I might, I've only found four types of Android specific startup opportunities, none of which have made me particularly excited about investing:

1. Alternatives to key OS functionality (think Swype for the keyboard or Fring for the dialer). Because apps on Android can replace key OS-provided functionality, there are companies trying to invent a better keyboard, a better telephony app, etc. I love Swype (it is super cool), but as a business you only need to look back 15 years at all of the folks who thought they could replace subpar Microsoft-provided functionality for Windows.

2. Enterprise-proofing Android. I think there might be a couple of companies who will build interesting acquisition targets here (especially in the crumbling of the RIM empire) but there is high risk that someone on the actual Android team may wake up soon to the fact that CIOs everywhere (who buy "in bulk") want smartphones made enterprise-safe and thus significantly shorten the runway for all of these startups.

3. Making Android a much better platform for other sensor input. Blood glucose monitors. Bluetooth video cameras. Traffic analyzers. All of these seem to me to be cases for Android's more general form multi-tasking which is quite a bit better than the mongrel Apple shipped, but in the fullness of time, I suspect the folks in Cupertino will catch on to this (or at least the batteries should catch up).

4. Customizing Android for dumb "box maker" OEMs. This is an interesting small business opportunity, but if you want to know what the limited upside is here, just go and find one of the Canonical guys who've been dealing with HP and Dell for years.

Other than that, Android is only better for startups for one Really Big Reason: it puts more HTML5 capable browsers in the pockets of regular users. But guess what? No matter who wins (iOS, Android or even Rim/Nokia), these very capable browsers are now jacks-to-enter in the realm of smartphones.

Smartphones are great for startups, full stop. But Android versus iOS? Sort of like wondering whether the early customers to Amazon.com in 1996 were coming from Windows or Mac. Either way, the big deal was access not whose vehicle you came in.

Comments Permalink