[historical post via archive.org ]  There are plenty of blog posts explaining fancy equations for how to divvy up ownership between founders based on weird things like, 5% goes to the idea creator or 10% goes to the CEO. They always seem totally arbitrary, but not because of the weird rules. It all comes down to the fact that measuring work is crazy difficult.

At Assembly, we have an even more complex problem: How do we evaluate individual tasks as a unit of ownership? In other words, what percentage of ownership should we give away for something, like designing a new logo or building a homepage?

I’ve always liked the idea of splitting ownership evenly between founders. Plus, when you include vesting schedules, everyone is rewarded based on time spent. But, we can’t use quite the same idea here, since, on Assembly, team members are more fluid and may not all work full-time on the product. Also, judging individual tasks in units of time never seems to work in practice.

Instead, we started with a rough estimation of what founders would receive for a years worth of work, guessed how many individual tasks they would complete in a year and used that percentage of ownership per task as a base value for each Assembly bounty. But, just like startups, early contributors should be rewarded proportionally more to match the risk involved. Right?

So, we used a sigmoidal curve to continuously reduce the value as the product matures until we hit a minimum value, which matches up with what a late stage employee would receive. If you’re interested in the exact math, check out Andrew Barisser’s explanation of the algorithm.

But, I know what you’re thinking. This is only a single estimation. What about bounties that are worth more than others? We took a look at how users were valuing bounties across all products and eventually settled on an exponential scale with that baseline used as the middle value. Here’s the final result.

For most products, one of the suggestions should fit each bounty. We even give examples of what different types of bounties would be worth, given the scale. But, for those bounties where you really want extra control over the value, you can switch to custom mode and enter an exact number of coins. And, if you want to award a custom amount of coins based on percentage of the product or based on financials, we have calculations shown for that as well.No matter how much fancy computer algorithm magic we applied to this problem, it was never going to be perfect. But I think we did a pretty good job of meeting our goal: giving users all the tools they need to confidently value a bounty.Try it out live and let us know what you think. Ping @vanstee or @abarisser on Twitter with any questions or feedback.

By Patrick Van Stee