News for hackers. Stuff that matters.
Re:Squarespace to Go Private in $6.9B All-Cash Transaction w... by dartos on Tuesday May 14, 2024 @08:15AM (#40354357)
> In contrast, people doing R&D in pharma or other similar fields will spend months or years working on a single project that doesn't go anywhere because it is too hard.

So does your argument boil down to “software moves faster than the sciences, so it’s not R&D?”

Initial development isn’t R&D. Like you said, setting up a webpage and hooking up payments isn’t R&D.

After a product is established, future work is R&D. A cycle of desired outcome, hypothesis for how to get there, and successive implementations until something sticks or the project fails and the hypothesis is proven wrong.

Re:Squarespace to Go Private in $6.9B All-Cash Transaction w... by pclmulqdq on Tuesday May 14, 2024 @05:59PM (#40360637)
No, it's not about things moving fast. It's about projects carrying specific kinds of risk and developing a specific kind of output. Namely technical risk and intellectual property.

Technical risk is the risk that your technical development won't pan out. I'm not referring to schedule risk - the risk that it will take longer than you thought - but actual risk that it will never work at all. Intellectual property is a durable, hard-to-copy advantage over your competition.

Pharma is a great example because it is a very pure form of R&D - you find a drug, test the shit out of it, those tests mostly go negative, and if you succeed, you have a formula and a manufacturing method that lets you charge extortionate rates. Most software work is not like this at all.

And no, ongoing future work is absolutely not guaranteed to be R&D. If anything, it is the opposite: most startups begin with R&D - creating an uncertain product with a (hopefully) durable advantage - and move out of doing R&D as they mature and the competition catches up. That is generally the lifecycle of companies that develop new product categories, from Singer (the first company to produce a sewing machine), to IBM, to Google.