A small enough community can achieve this level of surveilancesurveillance with simple observation
You need...
- An insecure leader willing to dominate the tribe with violence.
- Sycophantic members of the tribe willing to narc on their neighbors.
This could be achieved with the first chieftanchieftain-oriented tribal social structures, which occuredoccurred in the Neolithic period (10,000 BC).
However, this same behavior could just as easily be familial (a father tyranicallytyrannically watching over his brood), which means you can achieve your goal as early as 300,000 BC.
Unless...
You don't define what you mean by "constant surveillance." If by that you mean "somebody else has eyes and ears on you 24/7" then I predict it will come available sometime around year 2150. We can monitor communications, some viewing habits, some transit, and we have cameras in many places, but the reality is that we can't monitor the general population 24/7 today.
Conclusion
The goal could be achieved by...
- Family units as early as 300,000 BC (possibly earlier).
- Tribal units as early as 10,000 BC.
- Unless you really mean "complete surveillance," then it won't happen IMO until about 2150.
**Edit:**
A couple of commenters have suggested that we can achieve complete surveillance today. It's true that we have the technology to make cameras and microphones... but that's not actually the problem.
We have such a deluge of data right now that people are actually contemplating using magnetic tape to try and handle the data flow (Spectrum, IEEE, 09/2018) and complete surveillance would require increasing that data flow 10,000 fold (at least 10,000 fold. How many houses and businesses have cameras in every room today? Answer: almost none. [35.7M houses in the U.S. alone, what, average 7 rooms + garage per house? That's probably small... 300 million new cameras+mics, just in the houses... just in the U.S.... And that's just houses....).
People who think this is achievable with today's tech haven't thought the entire problem through. That data needs to be captured, transmitted, stored, evaluated... ugh (you'd need a third the country's population just to review all this data in a timely manner, the computational power to evaluate that much data realtime is appreciable). We're no where near the ability to handle that much second-by-second dataflow.