It won't work because people cannot do anything else with the money and thus will spend it on some project. They will not choose the best project as they don't have any particular incentive to do so. They will rather spend the money on the project of the people they are friends with.
To make it work I would propose the following changes:
- Employees have to invest real money (e.g., from their salaries). Investments are, of course, completely voluntarily. Investments could be limited to e.g., $1000 per employee per year. The important thing here is that the employee has to invest real money that she/he could use otherwise, so they will only do so for projects that they really believe in.
- IBM would increase the investment to e.g., 10x the value the employee invested. This would ensure that there is a boost to what an employee can achieve with her/his investment.
- The employee gets a bonus if the idea turns into a (viable) business project / product. The bonus could be e.g., 10x the value invested. This would ensure that the employees have an incentive to participate in such projects and that they really choose projects they think are viable rather than the projects of their friends. There could be additional factors, such as a 100x boost if the project not only succeeds but really takes off.
How to choose the "boost factor"? Well, if 10% of projects succeed on average, then the boost factor should be bigger than (1 / 10%) to ensure that investors get on average their investment back (and thus are motivated to participate).