[rbobby]

It's pretty difficult to get software development to qualify for SR&D. The way it was explained to me by a consultant was to qualify there had to be technological risk of failure (i.e. it is unknown whether or not it is possible to do task X via software).

So your average SAAS startup doesn't face technological risk, even though they face lots of business risks (might not be able to hire good enough developers, might not be able to produce bug free app, etc).

However the first few companies using deep learning to do facial recognition from photos (eg. facebook automatic tagging)... well they would have faced technological risk. It wasn't known at the outset that deep learning would produce sufficient quality matches to be useful.

On the downside... big organizations that can pay big bucks for assistance with writing SR&D applications can qualify for S&RD. Several of the big Canadian banks have recovered considerable amounts. Which is pretty outrageous on the face of things... how much actual SR&D and technological risk is being faced in a banking environment? None that I've ever seen (well... except for one project I was involved in... but that was a trivial "lets try this and see if we can use shared record locks as activity signals for a job scheduler").

In a way scientific/technological risk is a bad measure for software R&D. In general there is very little technological risk, and what risk there is is generally tackled via prototyping and/or early testing. The meat of the development costs comes much later... and there's no gov't tax support for that at all.

[thaumasiotes]

> The way it was explained to me by a consultant was to qualify there had to be technological risk of failure (i.e. it is unknown whether or not it is possible to do task X via software).

Ok...

> However the first few companies using deep learning to do facial recognition from photos (eg. facebook automatic tagging)... well they would have faced technological risk. It wasn't known at the outset that deep learning would produce sufficient quality matches to be useful.

This is a wildly different standard. It certainly was known at the outset that facial recognition from photographs is possible to do in software, because people do it all the time. They just didn't know how.

[stormbrew]

> It's pretty difficult to get software development to qualify for SR&D.

Every Canadian software company I've worked for in my 15 year career has used SR&ED.

Every single one. Most were small (ie. sub-25 people employed, most of those also making virtually no profit at all), a couple are large. Some tried to do it in house with only a lawyer to do late-stage massaging of the reports, others hired full on consultants for quite a lot of money to do it.

And they all got money out of SR&ED.

This is anecdata, but I don't think the idea that it's pretty difficult washes.

[herge]

I think banks face big risks with software projects through project management, but that's less tied to SR&ED ;)