@Clint Batman You are correct, the data does come from the consumer. Your question is geared toward the user interface and although we have ideas in this area, for the prize we are more concerned with the back end than the front end.
The data can come from a series of questions, that is tied to a minor decision tree. The data can come from using OCR from a scanned tax return and couple this with the questions, eventually we should be able to connect directly to bookkeeping software and/or tax preparation software. The way we do it today is get a copy of the tax return and ask a series of questions.
There are a few OCR programs that will help with auto filling tax returns. There are no proactive tax planning software packages on the market other than what we have internally.
For the sake of this prize, we are looking for a way to build the software so that the ultimate accuracy is higher than a human, which in this case should not be hard since very few humans that can apply the best of well over a hundred tax strategies, uniformly, consistently and without a bias toward ‘favorite’ strategies.
A good analogy is that we are focusing on the engine, the type of car it will go in hasn’t been determined yet. But, this is software for the future, so we need to “skate to where the puck is going”, as Wayne Gretzky famously said, and we know that is likely to be direct feeds from API’s or verbal question and answer. Not data entry. But for now, it’ll be data entry is some form.
Sorry for the long answer just wanted to be clear, hope I was.