Temperature is the most commonly measured process variable. Measuring it in most cases is trivial!
Show me an RPi project that calculates the temperature of a blast furnace by measuring the air pressure and the speed of sound in the vicinity and I'll change my mind.
What's distressing is that these folks will win in the end, and IoT wind up being quad-core devices @ 800Mhz because everyone is writing their system nodes in node.js or running a JVM. Small and tight just isn't worth the effort anymore.
Embedded hardware still has its place though. General purpose kernels like this can't handle timing-sensitive operations well, so there will always be a need for small and tight real-time programming. Other solutions exist such as binding a separate microcontroller to the host OS, but they're clunky and require additional drivers/libraries.
But you're right, IoT will end up being quad-core devices running some version of Linux because it's an easily accessible development platform that provides TCP/IP and a bazillion actively maintained drivers. And why not? The silicon is getting cheaper every day.
I would say the vast majority of projects on Github qualify as 'miscellaneous crap'. Most employers these days require a link to your Github profile during the application process, which can be a pain if you have been working with closed-source systems your entire career, or are relatively new to the field (Github isn't part of most CS curriculums). Throwing up a few toy projects can be enough to get you through the screening process.
With that said, I personally wouldn't submit a project like this to HN (I have a very similar project on GH). But if it got enough up votes to get on the front page, who are we to judge.
See the page for farmers -> http://farmhack.net/tools/fido-temperature-alarm-sends-text-...
Code here -> https://github.com/fidoserver/Fido
Blog post -> http://www.knightfoundation.org/blogs/knightblog/2015/2/10/k...
Project page -> http://farmhack.net/tools/open-pipe-kit