Fiverr Forum

Fiverr could allow sellers to view buyer requests in categories where they don't yet have a gig


It would be useful to be able to see buyer requests in categories where you don’t yet have a gig, so you can see what requests/types of requests are coming in in those categories and use that to decide whether to create a gig in those categories.

I would also like, the other people have asked, for there to be a flag option on buyer requests, mainly to flag requests that are sellers asking for work and not those requesting work done. It could also (assuming it doesn’t already) work like a spam filter where it gets trained to identify those that are really sellers promoting themselves and also identifies those that are valid buyer requests.


You’re able to do some kind of work in every category?

Flagging would be great. It needs to be sent to a human though. Otherwise it won’t work.


Probably not every category, but I would like the ability to look through each/any category to see what the buyer requests are, what they’re like and use that to determine whether it would be a good idea for me to add a gig in the category I’m viewing. I want to see the details of the requests before, not just after I create a gig in a particular category. It would let me know whether I would be able to do the sorts of requests being made in that category relatively easily and whether it would be a good idea (and profitable enough) to make the gig there.

But spam filters work. And machine learning works (generally). Basically the more it gets trained, the better it should be (if they were to have a system like that). It may get a few false positives (but they can be checked/moved manually), but the more it’s trained, the better it should be, and the fewer (lower percentage of) false positives it should get (ie. lower amount/percentage of wrongly classed buyer requests).


Machine learning only produces consistent output with no human input. Humans screw it up, invariably. That’s because humans don’t always think logically, and computers don’t think at all. Ever seen 2001? The problem with HAL is he was given two conflicting instructions, from less than logical humans.

Computers are great for mindless repeat tasks. They do not think.

Current machine learning consists of “try until you’re not wrong.” Machines have no intuition. (And who decides what is “right”?) Not only that, but with the current trend of ignoring real programming, everything’s going to get worse.

Spam filters don’t work though. At least mine don’t.