According to Stanford University, there can be merit in any idea. Using a technique they call rapid prototyping, a lot of different ideas can be generated in a short amount of time. This is done by forming 5-6 groups of three people each, with different professional backgrounds. Each group thinks of an idea, presents this idea to the other groups, and then receives feedback. Then, team members are shuffled around to form new 3-person groups, and the process repeats itself. This is done a couple of times, until a large amount of ideas has been generated.
During each idea generation round, groups should keep to the following schedule:
(0. Decide on a problem you want to address, e.g. hunger)
1. What are the target communities? (e.g. homeless people and bakeries)
2. What ‘tech’ do they use? (e.g. a shopping cart filled with rubbish)
3. What behaviour do you want to promote (e.g. bakeries give away left-over food to the homeless)
4. Create a prototype invention (e.g. a tracking device that can be attached to shopping carts, and starts to blink whenever there is left-over food somewhere nearby)
5. Measure impact (e.g. how many left-over food was collected by the homeless?)
6. Optimize (or pivot) & reapeat.
In the example I came up with a tracking device for homeless people. This is not likely a very good idea, but that doesn't matter. What matters is that there might still be some merit in this initially poor idea, that can be expanded upon by the other teams. By repeating the process, a lot of ideas are generated, and there are bound to be some good ones among them. The advantage of using the six steps outlined above, I think, is that they provide a guideline to come up with something that is practical and could - in theory - be immediately put into practice.