3 ms·

Its a thermostat, not a SR-71 or a fusion reactor. If the greatest engineers are required to work on thermostats, then we're doomed as a civilization. The qua

by VLM 11y ago

Its a thermostat, not a SR-71 or a fusion reactor. If the greatest engineers are required to work on thermostats, then we're doomed as a civilization. The quantity of great engineers will not be relevant to thermostats. Under employment is a thing now, and getting more serious over time, such that going forward it'll be even less of an issue.

vinceguidry 11y ago
I think you vastly underestimate the amount of work it takes to implement and maintain an API properly. It's probably more work than designing the thermostat itself was. Also the need to implement an API means you have to make the design of your thermostat API-friendly. So that gets harder too.
- TeMPOraL 11y ago
  It takes special amount of effort to design a device that is not hackable / modifiable by third parties. It's called tamper-proofing, and you do it on purpose, because even most basic engineering practices favour making things in a way that can be accessed, extended and hacked.
  - vinceguidry 11y ago
    Yes, we're discussing the continuum between tamper-proofed, normal hackability, and what my responders seem to want, a properly-maintained API. I think the normal run of things is just fine, I don't feel the need to demand every device connect to the network and listen on a port and implement a REST API. It would be nice, but it's not a must-have. If I need to hack a device, Google usually gets me there.
abakker 11y ago
Right - this is a very important point. If you're paying a lot for a thermostat, something is wrong. If you want this to be a smart device (as with many IoT functions), you're better off with an anduino and re-implementing thermostat function with a new array of sensors and software. an Arduino and a $15 PID will give you a better result anyway, and you'll also get to have the luxury of better sensor distribution, since you can instrument all of that stuff yourself. The major problem I have with IoT stuff, is that outside of the initial telemetry/sensor/data gathering phase, many of these devices won't change...The idea of a smart thermostat is not really special, nor is the app. Where the IoT would be best is in making it easy to diagnose heat loss in the system/air delivery inconsistency, and then provide a way to make physical changes to the system to improve things. Barring a way to act on the data, its useless. Its also useless to continually collect data over time, since the system has already been optimized. (unless you're expecting your instrumentation to either measure degradation of performance over time, or you can incorporate the instruments into a closed-loop system where they validate that the output remains constant).

by VLM 11y ago

vinceguidry 11y ago

I think you vastly underestimate the amount of work it takes to implement and maintain an API properly. It's probably more work than designing the thermostat itself was. Also the need to implement an API means you have to make the design of your thermostat API-friendly. So that gets harder too.

TeMPOraL 11y ago
It takes special amount of effort to design a device that is not hackable / modifiable by third parties. It's called tamper-proofing, and you do it on purpose, because even most basic engineering practices favour making things in a way that can be accessed, extended and hacked.
- vinceguidry 11y ago
  Yes, we're discussing the continuum between tamper-proofed, normal hackability, and what my responders seem to want, a properly-maintained API. I think the normal run of things is just fine, I don't feel the need to demand every device connect to the network and listen on a port and implement a REST API. It would be nice, but it's not a must-have. If I need to hack a device, Google usually gets me there.

abakker 11y ago

Right - this is a very important point. If you're paying a lot for a thermostat, something is wrong. If you want this to be a smart device (as with many IoT functions), you're better off with an anduino and re-implementing thermostat function with a new array of sensors and software. an Arduino and a $15 PID will give you a better result anyway, and you'll also get to have the luxury of better sensor distribution, since you can instrument all of that stuff yourself. The major problem I have with IoT stuff, is that outside of the initial telemetry/sensor/data gathering phase, many of these devices won't change...The idea of a smart thermostat is not really special, nor is the app. Where the IoT would be best is in making it easy to diagnose heat loss in the system/air delivery inconsistency, and then provide a way to make physical changes to the system to improve things. Barring a way to act on the data, its useless. Its also useless to continually collect data over time, since the system has already been optimized. (unless you're expecting your instrumentation to either measure degradation of performance over time, or you can incorporate the instruments into a closed-loop system where they validate that the output remains constant).