So taking 1 and 2 together: 1. Accept that you won't know what you're doing in 3 years. 2. Accept that you can't plan a successful startup 3 years in advance. 3. Start a small project, one you can launch in 6 months, and consider it a learning experience. Then, by the time your 3 years are over, you'll be ready to start a new startup, fulltime, having learnt a lot.

Do NOT plan a 3-year commitment in advance for a startup.

I think he may mean a web-based startup.

That would give you an edge when you decide to start.

or maybe your third or fourth project / experiment is quite interesting and you decide to have two or 3 six months iterations on it.

In that case you've found yourself your startup.

I'm not sure where I'll be in three years, but it does help to have a direction. I am married and we do want kids around the same time so that would obviously complicate things, which is partly why I'm trying to figure this out now. Again, thanks for the comments.

You can't count on having a perfect run. Experiment now while you still have more time on your hands.

That's why I'd suggest sticking with something a bit smaller and a bit more manageable that can launch early and grow slowly over time, which is much better-suited to your schedule and existing commitments.

For my startup, we initially thought the idea would take 2 weeks to do. In reality, it took closer to 3 months. In my case, once we started working on the idea, we discovered it was quite a bit more complex than we anticipated.

The way that I've figured out how to deal with this, is to punt absolutely everything that isn't absolutely necessary into the next 'version' after launch. Throw all of those cool features into your feature bucket, and only implement them once you've gotten the existing feature set working.

Your mileage may vary, of course. :)

My suggestion for you and any other person who wants to work on nights and weekends: above all else, ship something fast. Most people throw in the towel because they can't get into the positive feedback loop, it seems like it will take forever until they can get there, then ork gets busy and the baby gets sick and... and then they've forgotten about it.

Nothing replaces paying customers for giving ongoing motivation, feedback as to how to make your application better (via asking directly and reading the data about how they actually use it), and giving money to fund buying the things that you can't produce yourself (like, in your case, managed hosting).

Incidentally: although I'm definitely on the quick side of the bell curve, I took money for a product which was written (including the website to sell it and everything) in eight days. There is nothing wrong with taking six months to write something, but if you could have broken that into Phase 1: 1 month and Phase 2: 5 months you'd have money coming in, a website collecting links, customers giving suggestions, and MUCH less risk of building something no one wanted.

We're profitable but I still need a dayjob to pay the bills, so take my advice for what it's worth.

1. Shorter release schedule is always preferable. Like others have said 2 years to launch is a really bad idea. However, my question is: Where did you come up with these numbers? Do you have 2 different projects, or are you setting arbitrary timelines and want to fit a project to that timeline?

2. Going all-in is not a bad thing. I'm currently a "parallel entrepreneur" and it's very hard. It's much easier to focus on a single thing and hit it 100% rather than focus on a lot of things at 20%.

3. Sysadmin is easy: sign up for a service where someone else handles it. As a Ruby on Rails guy, I heavily recommend Blue Box hosting. It's like having a sysadmin on the team, and it's not much more expensive than Slicehost. It's a good skill to learn, but I'm with you: let someone else do that stuff.

4. It sounds like you need an idea. I know everyone says it's the execution, not the idea, but you need _some_ idea to execute on. I would suggest using your time now to try and find 2 things: an idea and a co-founder. Going to local meetups, talking to people, and such would be good things to do at this stage. Find something you can be passionate about, or else you'll never find the motivation to spend your nights & weekends on it.

My original thought process was that it'd be better to pick a long term ambitious project and launch a minimum viable product for it in two years (not a full project like I think I implied in the original post). As you and others pointed out though, I should shrink that even more until I have something that I can release in months, not years.

I have plenty of ideas, but they're mostly, in pg's words, science projects, not company ideas. See, for example, HNTrends.com. Neat idea, but not exactly something that's going to put food on the table.

There's pros and cons to looking for a big idea. If I identify something that I'm passionate about, execute it well, and lady luck is on my side, I'll be in good shape. But, the longer I drag my feet looking for a big idea, the more I increase the chances that I'll miss out by not simply executing a small idea and gradually growing it into something big. All that said, I think your initial direction matters a lot. I imagine many founders have had great ideas that they wish they could execute but they can't because they've committed to something else which is why I'm taking the time right now to figure things out.

I suppose what it all boils down to doing exactly what you said: find something you're passionate about. that'll take care of almost everything else.

And passion definitely won't take care of everything, but it will solve the motivation problem, which I think is the biggest problem that a nights & weekends company has. You still need to find customers and all that, and while passion helps, so too does having a useful, viable product.

I would suggest trying to get outside of the HN geek echo chamber. Ignore Twitter, the iPhone, and all the other hot stuff right now and try to find an area or group that's totally underserved on the web. Ask friends, girlfriends, family, etc. for ideas on things they hate about the Web. In my case, my wife found that small plant nurseries are stuck in like 1996 when it comes to the web, so we started http://doleaf.com Sure, it's not super-sexy, but it solves a real problem for these people, and they're willing to pay for it.

Why not have both? What I mean is, take a big project, break it into smaller pieces of 3 months each (features 1-10 in first iteration, 11-20 in the second and so on). Since you'll be releasing early, you can gauge the response to your app, and adjust the next iteration according to the feedback (or even drop the project completely if needed)

Every year I wanted to create something I called a cyber lemonade stand. Tim Ferris calls these muse businesses. I call it something simple where I charge money, people pay me, and I learn that I can do it. This is a big confidence builder and something you can start doing right away. http://www.feedbackarmy.com was my cyber lemonade stand, After the Deadline http://www.afterthedeadline.com was my startup.

What I find interesting is that I wasn't willing to take the steps towards the lemonade stand until I was out of the Air Force and had no job. If you're serious about starting up though, I really recommend you start on this concept right away. A small project will let you experiment with the things necessary to get something off the ground (contacting bloggers to generate PR, hiring out small stuff you can't do well yourself, etc).

Once you hit that magical time when you can startup--go for it. But you're thinking the right way trying to get some training in before the big race.

learn more about the things you need to know when they come up. you don't need full-time availabitlity to make a startup happen.

lets you make sure you're doing what your users want, instead of what you think they want. build and improve with their feedback in mind.

Everything you mention in this paragraph about "business knowledge" falls into my category, "Details to worry about later."

You omit, however, the single most important business issue of all, the only one of concern for now, and in fact, one the you must be concerned about now: consumers. Whether you call them customers or users, someone will be using your software, right? Then you better find a way to get to know them now. This is a critical component of application development, every bit as important as the technology.

Lots of things about your consumers could change between now and deployment, but you have plenty of time to incorporate them into your plan. So make sure you do.

I think Twitter is kind of stupid too.