Thanks for the idea. I have considered open source JUnit Max. However, it would require significant investment on my part (perhaps a month of work before it is ready to be shared) and significant on-going investment (open source projects need to be nurtured to live), all with no conceivable monetary payoff. It doesn’t make sense. It’s a pity that the code will die, but I hope that ideas of test prioritization and in-place feedback will be taken up by other IDE developers.

Thanks for the thoughts. I did consider the freemium model. A freemium business needs to offer:
1) Value “for free” (that is, in exchange for users investing time but not money)
2) A “conversion moment”, where the obviously right thing to do is pay for even more value. The “15 SMS messages a day free” iPhone apps are like that. You’ve already sent 15 messages that day. You really want to send the 16th. $9.99? No problem.

#1 was easy. Around half of the people who tried Max really liked it (modulo scaling to large workspaces, which could have been solved with more engineering). #2, however, I never really figured out. What would be a good dividing line–# test runs/day? #tests/day? #tests total ever? I never found a “conversion moment” that was convincing to me.

Dividing “personal” and “enterprise” makes good sense. Again, though, I never figured out what the “enterprise” edition would include that would motivate a manager to buy it in his own self-interest.

Underpinning all of this is my suspicion that not that many Java programmers rely on tests for feedback minute-by-minute. Some, yes, but not a market’s worth.

In my opinion Max is sufficiently different from JUnit to require “user education” and community acceptance, just as JUnit did some moons ago. A free (beer) or open source entry level implementation would in my opinion build a user base and extend mindshare, paving the way for an advanced for-pay “enterprise” version. Or am I wrong?

I would like to hear more about how our perspectives differ. I do plan on developing tools in the future. I don’t like the uncertainty of product development, but it seems to offer me the best chance of being able to program for a living and live somewhere beautiful. Two lessons: sell to companies instead of individuals & include a service component from day 1.

I agree that marketing for early-stage products is different than for more mature products. The goal in both cases is creating demand. In the early stages marketing creates demand for the purposes of generating seed capital and sizing the market. Too much demand, or demand from the wrong kind of customers, is as damaging to a product as too little. In later stages marketing creates demand in existing markets or opens a product to new markets.

The conundrum I faced was how to market without any cash. I do have my reputation–people will (briefly) listen to what I say. That’s why I used the media I used. Actually, if I had to do it over again I would attach my name less prominently to the product. Some people bought Max because it was a tool I produced, not because it was a tool they really thought they needed, and that delayed clear feedback. The signal that clinched the decision to deadpool Max was the lack of word-of-mouth. Subscribers were telling their friends, but their friends weren’t buying.

Regarding patience, I’ve seen products take a long time before they took off. As a penurious bootstrap startup, I don’t have time for that. I can’t afford the opportunity cost. Because Eclipse is so expensive to develop for, I couldn’t just let Max simmer. I needed several months of engineering to make Max solid and performant. Without early revenue to pay for that, I needed to move on.

IMHO, any new product requires a LOT of patience and time before people start to accept it, talk about it and buy it. Behind every overnight sensation is a lot of hard work and time. Bonfires may happen but they never last long unless you keep feeding them…