I’m one of the cofounders at data.world and you definitely make good points… We encourage all users to post a license with their datasets but, like open source,

I’m one of the cofounders at data.world and you definitely make good points… We encourage all users to post a license with their datasets but, like open source, not everyone will maintain or honor these. This is definitely something that we feel we can help the open data community with and encourage even more. (As an aside, in general you should note is that a lot of data scraped from websites is something that you’ll need to be careful with). When it comes to data quality, the world is a messy place and the data that comes from it is messy too. Most professional data scientists spend an inordinate amount of time cleaning datasets, doing feature engineering etc… it’s part of their job description. We’re trying to eliminate some of that repetitive work by making sure that people can comment on, contribute to and give some signal back on the quality of the dataset. We also think a dataset is more than just the data... on data.world you can upload code, Notebooks, images, etc... anything that helps add context to the data. Finally, when it comes to size... ML definitely needs it. However, there's a lot of interesting data out there thats still very complicated, very useful but not that big. Most datasets in the world are well under the terabyte size (or even 100s of GBs in size). We're rapidly expanding the size of datasets we support because we want that stuff too but we really want to help people understand all the data in the world!