I hate to break it to you, but virtually everything at this scale is an estimate based on models and not direct data. We don't e.g. get more reliable GDP number

I hate to break it to you, but virtually everything at this scale is an estimate based on models and not direct data. We don't e.g. get more reliable GDP numbers because we get hard data; we get it by creating new models that recognize and/or emphasize variables which were previously being ignored. The reason to be skeptical here is not because "it's an estimate"; it's because it's one group's model, studying a very nascent space for which there's been very little academic research. What this estimate shows is that this is a clearly very large economy that isn't being rigorously modeled or studied right now. And we should probably devote some resources to that since what happens in this economy will meaningfully contribute to employment and national GDPs.