Scientific Atlanta… there’s a name I haven’t heard in a long time. Didn’t they use to make crappy cable boxes, back when cable TV meant a box that connected to the antenna input via coax.
One other thing that I think feeds into these acquisition mishaps is that Cisco has, in my opinion, consistently over-estimated how much intelligence would be needed (or wanted) in the core network. In their view, intelligent network services = expensive network devices = revenue for Cisco. I think what the Internet specifically and IP in general, as well as the evolution of LAN technologies over time have proven is that when it comes to the core network, simple is almost always better and intelligence should move to the edge, where innovation can happen quicker and where services can be implemented in software.
As an example, at one point they had what was, essentially, a middleware system (like Websphere,) which they called Application Oriented Networking. The idea was you would deploy these on your network gear, throughout your network, and it would provide message routing and translation services. They had a whole "architecture" built for it, called Services Oriented Network Architecture[0]. I don't think the people who built it really understood that it provided no real advantage over a cluster of middleware/ESB/MQ servers in a data center and that nobody was going to pay a huge premium to build that capability in their IP routers.
0 - https://www.cisco.com/c/dam/global/it_it/solutions/ent/tecno...
Ironically, those set top makers were in a perfect position to take advantage of it. They could have been Roku - they already had huge market penetration.