From the candidate perspective, one way to assess the viability of the hiring strategy is imagine if most, or a significant number, of companies doing this. This eats significantly into each candidate's time, and edges the job seeking dynamic in favor of the companies, and against the candidate since the candidate is forced to value the cost of opportunity searching versus doing something else. This also dramatically would increase the stress of a job search - I experienced it first hand over the course of last year when a good portion of companies wanted projects done, all which would have totalled maybe 100 hours of extra work on top of everything else normally associated with interviewing.
This is not a sustainable balance for job seekers - ultimately this acts against most job seekers' self-interest. I'm fine with doing project-based tests ability/attitude-wise but this reasoning is why I pass up on them.