Automated Impact Evaluation & Marketplace Pricing
- “The goal is definitely to create an open market. However, it's not like two people will be selling the exact same dataset. In fact, with IFPS CID's we can ensure that two identical datasets don't exist.
- I'm imagining there will be different types or classes of data, within which there will be a range of prices for similar datasets. One easy example to visualize is genome data.
- Prices could be determined algorithmically depending on the size of the genome data file, type of sequencer used, data quality scores, etc. without ever explicitly factoring those specific metadata fields into account.
- Idea: The deal-making algorithm could fluctuate the price of the dataset
- The dataset price makes it so it’s like it's sitting in an order book next to similar datasets. When data consumers want to train a model or something on a specific type of data, they could even filter by the metadata parameters they care about, then choose to pay up to a certain amount depending on what the shape of the order book is.”
- Opt-in: users to choose whether to enable programmatic deal-making or not.
- People could initially even choose to list a data asset at a set price then enable algorithmic price finding later (e.g. we could even show a calculation for the current average price of similar datasets before someone chooses to click publish).