When people start talking about crazy sci-fi scenarios about "the machines taking over", and I try to ponder realistic ways something like that could actually happen, my speculative journey currently leads me to the future of farming automation.
Maybe in 20 years Farmlogs 2035 will be powering fully automated farming that will be beyond our current imagination in efficiency. Application of machine learning and evolutionary programming will lead to it doing things with seeds and crop planting patterns and watering frequency that will be outside of our ability to comprehend.
Because of this amazing improvement in efficiency, we will need much less land to support the population. Fields would lie fallow and be turned into condos. It is at this point that we will be past the point of no return. Further into the future, if Farmlogs AI 2075 just decides that it has other plans than feeding humans... or even if there was some weird "bug" that caused a severe downturn in crop yields for a year. We would have to scramble to grow enough food without the incomprehensible techniques used by the machines.
I don't think that is at all a likely scenario. What I would want to see is a policy that they won't sell or share the data they collect. If they have one I haven't been able to find it.
Farm equipment companies and chemical manufacturers have tried to offer this service before and failed. Partially it was way too early, but the main thing was farmers didn't trust them with all their data.
I think FarmLogs has the potential to be another YC unicorn if they stick it out and aren't acquired. Trust me if John Deere or Monsanto were to acquire them all hell would break loose.
All Hell wouldn't break loose, because FarmLogs isn't unique. Deere and Monsanto each have their own "cloud-based" farm management products: Deere with myjohndeere.com, and Monsanto with the Climate Corp umbrella (including Precision Planting, FieldScripts, etc).
My fine employer is Climate Corporation, owned by Monsanto. We have worked on - I dare to say led even?, but I was not involved - an Open Ag data consortium: http://openag.io
> Maybe in 20 years Farmlogs 2035 will be powering fully automated farming that will be beyond our current imagination in efficiency. Application of machine learning and evolutionary programming will lead to it doing things with seeds and crop planting patterns and watering frequency that will be outside of our ability to comprehend.
My current knowledge is that we're looking to do that kind of thing: with the general goal of increasing sustainability and feeding people more. My hope is that more efficient farming, combined with careful attention to urbanist approaches, will actually allow reclaiming more land for land trusts & national forests.
I don't get what is so impressive about what is essentially an accounting app. it's a well done technology but some people get so carried away by this farm tech stuff... a CRUD app is not going to make farms more productive... I've
seen that farmlogs is using satellite NDVI data now to predict yields etc.... I've done a data science project and found the the NDVI has minimal correlation to yield. part of we thinks that this is a group of CS people pretending to be ag engineers. ..
I think that was the point of the podcast. It started as a typical and shitty CakePHP / Backbone CRUD app like everything else. You don't need much to get started.
There's a lot more in there now, but it's still built around boring record keeping software. I wouldn't even say accounting software.
We're ingesting shitloads of geospatial data specific to each field from government and third party sources, from weather data like historical rainfall and temperature, to in season and historical satellite imagery, to sensor readings from their own combines or smartphones.
This is going to help farmers understand their fields and how their crops are growing better and soon allow us to provide agronomic services like variable rate prescriptions. We're not predicting yield from NDVI. There is a correlation there, but like you said it's moderate.
I think there's about a dozen startups in Australia doing something similar. Sort of makes me wonder why Farm Logs is getting so much press coverage recently - I can only suppose that being in the US and having done YC is a huge boost.
Maybe in 20 years Farmlogs 2035 will be powering fully automated farming that will be beyond our current imagination in efficiency. Application of machine learning and evolutionary programming will lead to it doing things with seeds and crop planting patterns and watering frequency that will be outside of our ability to comprehend.
Because of this amazing improvement in efficiency, we will need much less land to support the population. Fields would lie fallow and be turned into condos. It is at this point that we will be past the point of no return. Further into the future, if Farmlogs AI 2075 just decides that it has other plans than feeding humans... or even if there was some weird "bug" that caused a severe downturn in crop yields for a year. We would have to scramble to grow enough food without the incomprehensible techniques used by the machines.
Ok... getting back to work now.