FarmLogs | Hacker News Reader

FarmLogs

blog.samaltman.com

444 pointsclschnei87 comments

The problem with FarmLogs is the same problem with anyone wanting to provide cloud-based services to agriculture: who owns the data? And specifically, FarmLogs Terms of Service clause 7 "Content" does not leave one feeling warm and fuzzy on this topic.

This has become more of an issue with farmers over the past year, as they realize their data is a commodity just like what they raise--whether crops or livestock. How much of an issue? Two examples.

First, the Iowa Power Farming show was last week. One of the largest ag shows in the nation, and the back of the program had an advertisement from Ag Leader which started with the line: "Settling for a precision farming partner that wants control of your data just doesn't cut it." Additionally, several vendors in the precision ag and sensor/drone areas were stating similar lines. Whether BS or not, they wouldn't be saying it if there wasn't a need for them to.

Second, the January 17, 2015, edition of Iowa Farmer Today also addressed this issue in their editorial page article "Who own precision technology data?".

It's not that I'm against using technology for ag. It's I'm against others using my data about my farming operations AND MY LAND and making money off of it without cutting me in--or selling/providing it (raw or as a "derivative work") to other entities I may not want to have it.

"... This has become more of an issue with farmers over the past year, as they realize their data is a commodity just like what they raise--whether crops or livestock. ..."

What is stopping a local server/you own the data solution? Is it the cloud solution is the only one on offer or is it too hard to get the hardware/software to the gate?

Couldn't you say that about most tech solutions available today?

Particular as a farmer I imagine you have little time to be screwing around with a local server that's prone to failure. A company looking after that for you is probably quite tempting.

> you have little time to be screwing around with a local server that's prone to failure

While I agree with the fact you shouldn't be screwing around with a server, that doesn't mean running your own server with your own data on it isn't a good idea. It's just we're not quite there yet!

I've got no experience/knowledge of the FarmLogs platform, but I would guess the reason farmers arn't doing this themselves is probably the same reason they don't host their own email platform: It requires a level of technical skill, plus time and money to do so so that it works well.

I looked at it due it's consolidation of weather records with fields. Not totally impressed with the implementation but there could be significant value in analysis of larger areas, catching disease spread early or tracking pest migration.

This article says the equipment vendors are going out of their way to lock in the farmers to specific platforms and making it very hard for farmers to do anything without the manufacturers/dealers.

http://www.wired.com/2015/02/new-high-tech-farm-equipment-ni...

There are desktop-based farm management solutions out there from companies like Ag Leader, John Deere, Trimble/Farm Works, and others. Some of these interact with smartphones or tablets (sync on LAN, maybe Internet).

Many of these companies are also moving towards cloud/SaaS-based solutions and developing private networks to facilitate real-time data ingestion from their equipment (tractors, combines, planters, sprayers, irrigation systems, wagons, etc).

So there are local solutions, and they're installs like any other software.

It's usually too hard to set up such a system yourself.

I'm strongly rooting for https://sandstorm.io/ here - I hope they'll be able to provide a good alternative for SaaS businesses.

Great point. I'm one of the founders of FarmLogs and I can tell you we started this company to help make farmers (like my family) more successful. We don't have ulterior motives with regard to using farmer data. We help them use it to be more profitable and that is all.

That said, it looks like we aren't doing a good enough job making it obvious in our ToS. That page has not been updated in a very long time. I am going to be sure to update it to more accurately reflect our values.

Kudos, this is exactly the kind of response I love to see.

Good stance, however, that leaves unaddressed two key prongs of this issue:

1) Is this actually a concern to the people you want to be selling to, as evidenced by talking to a number of them? Don't bother with surveys, just straight-up have a non-sales conversation about it with 10 people who match your target customer model.

2) Lots of people don't read a ToS. Or if they do, they're not going to be 100% confident that their interpretation is what will stand up. If this is a legitimate concern for your users, you need to allay that concern, directly, in your primary sales contexts. Make it a focus item on your landing page, with design such that people will see it and go "oh, okay, I don't need to have that concern after all."

As somebody who previously prepared many tax returns for farmers, it's great to see people trying and help them to be more profitable. A very small percent of them actually are profitable. (Not including the ones who are trying to show a loss for tax reasons :p) 20% is a good start, and hopefully tools like this will help more farmers, because I like fresh and real food, usually local, and many of them definitely struggle.

"...We don't have ulterior motives with regard to using farmer data. We help them use it to be more profitable and that is all."

Until the investors/board say otherwise. :)

Been there. Experienced it. Resigned rather than comply.

Understand FarmLogs is the whipping boy here because you're the thread's subject (and your TOS clause 7 really is scary), but my concerns are also with the others, including for example Deere.

And as your TOS are updated, things might want to consider:

1) Who owns the data--the landowner or the tenant? Which data specifically? Who can order it removed? 2) What happens when a farm changes owners and the new owner wants all data regarding his land removed? 3) What happens when the data is subpeonaed in a clean water lawsuit for agricultural nitrate runoff?

> Been there. Experienced it. Resigned rather than comply.

Not to derail the topic but I'd love to hear that story. Have you published it anywhere?

I second that, I'm intrigued.

/s/cloud-based services to agriculture/cloud-based services

http://openag.io/

It's good to see that consumers are beginning to question who owns and controls data that is collected from them. I sometimes think that the "big data" revolution is worse than anything the NSA is doing.

This is an excellent point. During university, I was part of a team that published a business case centred around the story of Vidalia Sweet Onions. This guy had developed a technology that took a sample of onions from a farmer's field, crushed the onions, and analyzed them for the characteristics that made for sweetness. The batches were tagged with grid numbers to enable farmers to confirm which plots of land had the type of soil necessary to make truly sweet onions. This then enabled the farmers to guarantee the sweetness level of various onions. The old guard was absolutely against it, as the collective of farmers that sold under the brand of Vidalia onions partly depended on the mysticism of not knowing which plots of land were able to generate sweet onions. Without this testing technology, the farmers were all on a level playing field, irregardless of their actual soil composition. The testing technology was able to clarify which emperors had no clothes. Whether a farmer can or should be able to own the data that is critical to his success is a question people never really had to consider before.

I liken it to mining. The companies who dig for various mineral deposits make significant investments into exploration of the ground. Their exploration generates a ton of data. They use that data to determine where to dig further and estimate the financial potential of the deposits. This then determines company stock values. It's ludicrous that such data may not be owned by the mining company that generated it (or acquired the rights for it). Obviously, what's happening with farms and FarmLogs is not a perfect analogy. But the parallels are interesting.

> And importantly, they started out doing something that any two programmers could have done.

Which is funny because many VCs in SV ask the question (explicitly or implicitly) "is this something two Stanford CS students can do?" If the answer is yes, they move along.

I want more FarmLogs stories to exist.

This is a silly question to ask, and I don't think most people ask it (or at least, not in this way). For almost anything any team can do, you have to assume there will be five other teams, just as good or better, doing the same exact thing.

So the right question to ask is "what is the distribution edge of this particular team?" For example, if you're first to market and have incredibly high growth, there is a good chance it will be impossible for other teams to catch up.

Product complexity is almost never a competitive advantage, and I think most VCs understand this much better than you give them credit for.

I agree with your argument, but I do want to point out (since I think people here sometimes forget this) that the "5 other teams" thing starts to break down once you leave the traditional Silicon Valley product domains.

When working in VC, it was refreshing to talk with someone not doing social/mobile/local/consumer. Once you branch out to industries like Mining, Oil&Gas, Agriculture, Automotive, Manufacturing, Logistics, Insurance, etc. just having 2 bright software engineers working on a problem at all was often enough to provide an edge.

Insurance is a good example. That industry move $2T every year in the US, which makes the US ad industry look tiny. However, it only sees maybe 1/20th the number of SV teams.

> For almost anything any team can do, you have to assume there will be five other teams, just as good or better, doing the same exact thing.

That may be true for your typical cloud / SaaS / consumer internet / AirBnb for XYZ -kind of startup but there is many many fields in technology where killer teams are exceptionally rare.

Most of these technology fields are not as sexy as building the next Dropbox but very often all the more sophisticated in their core tech. Think about bio-tech, energy, nanotech, lasers, (space) flight, AI, etc. It is never bad for a startup to assume that there is competition and to research it but it is very possible for the right people in a particular field to team up and be the best.

> Think about bio-tech, energy, nanotech, lasers, (space) flight, AI, etc.

I'm not sure. I tend to assume the market is efficient and I found that far more often than not, the assumption proves correct. I don't know very much about most of the fields you mentioned, but think of space flight -- multiple space flight companies were founded around the same time as SpaceX (including John Carmack's now defunct space company, Virgin Galactic, Orbital Sciences, Blue Origin, Planetary Resources, and probably many others I'm missing).

I think that as a rule of thumb, the startup market is much more efficient than it originally seems, even in very deeply technical, capital intensive fields.

I don't think this is true at all which is why you have more startups. Also you can have multiple startups doing the same thing across say different geographies.

The problem with VCs is that they understand a lot, but sometimes to ask silly questions in search of intelligence. In the worst cases they verbalize a "no" as the silly question.

The one counter thing I'll say to this is that Brad and Jessie both had some first hand experience of farming (Full Disclosure: I was in the W12 batch with them and talked to them both a fair amount during YC) They weren't just two software guys with an idea, they also had seen some of the pains firsthand and understood the market well. I think their unique backgrounds lent them to making this the perfect opportunity for them to tackle and ultimately one of the majors reasons they've made such a large impact so quickly.

That's a very good point - and despite my earlier comment just knowing where and how much the pain points are matters.

For example in the (UK) building trade there is a booming trade in "builder remediation" - fixing the accidental below spec work they did not notice. It can wipe out much of the profit a builder makes having to go in and pull a wall down.

I keep calling the solution to this as "bringing the factory to the field". That is measurements and control processes we see in the factory applied in less controlled environments. With mobile tech for example one could slap QR stickers on timbers, fittings and hard boards and at each stage of the build require a photo of the materials in-situ before they get covered in plaster or the next layer up.

This would help track an enormous number of issues in the building trade.

I can see how just a little inside knowledge can help.

It's worth pointing out that he says,

"And importantly, they started out doing something that any two programmers (with domain expertise in their market) could have done."

and the key part of that is the "domain expertise in their market."

Our socioeconomic system very carefully prevents most students with farming domain expertise from ever getting near the Stanford CS Dept. It seems like most of the value add here is from the domain knowledge, not from the technology.

Indeed. Two Stanford CS students could have created Google ... and they did! There's a lot that two programmers can do.

As an early user of FarmLogs (and as a programmer), I found it really useful. All I wanted to do was find out average rainfall on my land and make a way to denote the prime spots for crop planting on the land (purely as a hobby). It was super useful and intuitive for that, something that I couldn't get to make happen easily trying all sorts of things with Google Maps, various mapping APIs, and Google Earth.

Just curious, did you try ClimateBasic from Monsanto/ClimateCorp and compare some data for "ground truth"?

http://climate.com/products/climate-basic/

Climate didn't have this service available until recently. Anyway, they're probably using the same rainfall data (URD and/or AHPS)

As another farmer and programmer, who isn't based in the US, I have found all of the most desirable features (like rainfall data) are not available. I have tried it out, but there is nothing that keeps me coming back. 20% of American row crop farmers is still fairly impressive, but it will be interesting to see how they try to approach the rest of world.

Incredible. How do they do distribution? Quite a feat to reach so many farmers in 3 years.

farmers talk to each other, a lot. Ideas spread very quickly, word of mouth is huge.

Heck, just look up how combines travel from south to north during harvest. Farmers are connected and talkative.

Damn; I wouldn't have thought farmers would be a good group to target software towards, but it actually makes a lot of sense. Out of curiosity, I wonder how these guys actually manage to get those millions of dollars in savings.

Nice to see a new startup is trying and succeeding to make a positive difference for some pretty critical customers :)

Going to keep an eye on this group for sure. Something tells me they'll be around for a while.

>Out of curiosity, I wonder how these guys actually manage to get those millions of dollars in savings.

I'm guessing a big chunk of that figure is derived from a boost in crop yield due to optimal placement. So the bottom line cost for the farmer was similar but money went much further.

Reading the app description, more consistent data about fields + better budgeting towards fields would likely do it.

As an example, corn and soybeans are commonly rotated to help reduce the usage of nitrogen-based fertilizers in fields (soybeans put nitrogen in the ground, corn takes it out). If you have better data about nitrogen fixation, you can know if/when you should swap, and/or run the math on expected prices for soybeans vs. corn, irrigation costs, etc. Normally this is done in spreadsheets, but in the app you can get estimates and budgets for all of this stuff much faster (at least from what I read).

Farming is hard, and better support for data/analytics on farms is awesome.

You'd be surprised - a good friend of mine from high school was from a farming family, and they had a PC in the early 90's when hardly anyone had a PC in rural areas except schools and libraries. (I grew up in northeast Arkansas) They used it for accounting and taxes mostly (and we used it for games!) and they wrote the cost of it off each year.

The other major early infusion of tech into farming I can think of is the growth of GPS guided precision application in the early 00's. I was in college at the time and my dad worked for a farm services company; they had people that went out and drove around fields with equipment figuring out exactly where to spray the right stuff. I recall it being a big deal because it saved the farmers a lot of money on chemicals.

Likewise--my mother was doing our farm's books on an IBM 8088 back in the mid-80s.

I have my doubts about the figure quoted "20% of row crop farms." I wonder how this number was presented / justified with what data at the meeting?

Maybe something like 20% of row crop farms of acreage greater than X where X is fairly large?

The software revolution is making it possible to create world-changing companies relatively quickly and with relatively modest resources.

I wish there were a bit more meat as to HOW they did this. Or a link to a meatier write-up of the path they took from idea to present.

I've chatted with Jesse on a couple of occasions. The direction the company is going with big data and analytics really is quite interesting and they are taking an interesting approach to a real world problem.

It's interesting to read updates about them as they pop up.

I worked on a sugarcane farm management company a while ago and also built a backbone.js frontend site that tracked sugar cane harvesters and estimated the "outcome" to send trucks to offload it on demand with some colleagues (closed source unfortunately) in resume I like this area very much.

When I saw this post I went running to the Jobs page, and there my heart got smashed when I saw Clojure on backend positions... I am a plain Java dev with learning skills (worked with Groovy/Python/JS a while ago) but I definitely don't like Clojure.

Good luck guys

It's Clojure not Closure. Why don't you like it?

> I am a plain Java dev with learning skills (worked with Groovy/Python/JS a while ago)

Perhaps he's promoting some other JVM language -- he did mention Java and Groovy. If you type "Groovy" into HN search, you'll notice more than the usual amount of Groovy stories being submitted over the past week, and the recent comments mentioning "Groovy" are scattered around different submissions rather than clustered as usual. Groovy has more than doubled its percentage on Tiobe between January and February (if you get what I mean), and its project manager has just finished running a campaign among his Twitter followers to double its Github stars from under 600 a month ago. Because the Groovy and Grails project managers are now competing for their funding, expect to see more comments of that nature.

I think that's cynical (if I read it correctly), but maybe true? When money gets involved all kinds of weird things happen.

Rich Hickey is on the record as saying that he doesn't want to promote Clojure, and if people like it they like it. He'd rather have people who came to it voluntarily than who were marketed to.

I would like to know real answers to why people don't like it. I've used a (normal) number of languages over the years but I think Clojure is by far the best for all kinds of reasons.

It's interesting to talk to people who have gone as far as trying it but found they preferred something else. I can understand you might be put off from a distance by the parens, or the immutability or whatever, but opinions change (or don't) once you actually try something.

I used Clojure a lot as a hobbyist language for 2 years, then changed to another one. I found I missed the immutability the most, even though I hardly wrote multi-threaded code. Once I was thinking "concurrently", going back to having to think about whether a list or map was in the right state took some getting used to.

As for what I couldn't get used to in Clojure, even after 2 years the circumfix parentheses still seemed unnatural, even though I know they're necessary for macros. I guess using f(x,y) notation in school and one's first programming language is a bond too hard to break.

Oh shit what a conspiracy theory...

I played with Scala too... Maybe I am receiving a paycheck from all other langs besides Clojure to kill the damn thing.

Come on...

I just don't like the LISP like syntax and the general idea of it and I think that other languages like can provide the same nice pieces it does... Simple... Simple... No aliens, JFK Zombie killers, Umbrela corps or anything.

Thanks I've corrected it.

Lisp like syntax all those (((())))))(()() drive me crazy and I think Scala has the features of Clojure that I like without the LISPness.

But I have nothing against people using I just don't like it.

I thought you'd say something like that. Have you counted how many brackets though? It's often very close.

    (defn blub-extra [a b]
        (blub (inc a) (inc b)))

8 parens + 2 brackets

    def blubExtra(a: Int, b: Int): Int {
       blub(inc(a), inc(b))
    }

8 parens + 2 braces

    Integer blubExtra(Integer a, Integer b) {
        return blub(inc(a), inc(b));
    }

8 parens + 2 braces

It's roughly the same numbers of brackets (or equivalent) in Clojure, Scala and Java. They're just in slightly different places.

(excuse syntax I've not written the last two for years)

Thanks for the answer...

I guess my LISP Teacher ruined the syntax for me. Terrible terrible class.

Maybe you are right. I might give it a try again and see if my position still stand :D

Argument lists are in brackets rather than parens, which helps slightly with the visuals.

This type of story is so motivating to build vertical software, but for people like me who are trying to identify a vertical software opportunity, it's discouraging that these developers needed to spend a couple years in the industry to truly understand the pain point behind FarmLogs.

Does anyone have thoughts / strategies around identifying these opportunities, ideally without the need to spend years in the target industry?

Co-founder(s) with domain expertise?

There is no free lunch.

* Your hobbies and interest groups

* Your job (harder if it's just building tools for web developers)

* Your family members and friends' jobs & hobbies

* Basically anyone willing to talk about their problems with you

Unfortunately, you need to gain a lot of tacit knowledge about an industry to really solve the right problems for it. That takes time and dedication.

There are some ways to speed it up by talking to lots of experienced people, reading analyst papers, researching, or even shadowing a person in the industry for a week. However, nothing beats first hand knowledge.

I think you get at a great point that the onky way to solve the right problems is to be committed to learning about an industry through all available information channels. To be comfortable spending that much time on one specific industry though I think means you need to come in with a thesis to begin with, which to avoid circular idea justification means you probably need to come into it with a broader, more abstract thesis independent of industry knowledge.

The company seems very cool. One off topic thing that I can't help but note - just loading their homepage requires 113 HTTP requests. It's crazy heavy with tracking and designy tags.

http://tools.pingdom.com/fpt/#!/ZMQov/https://farmlogs.com/

@briholt, thanks for noticing that. We really appreciate you telling us about it. We have noted it and our designer is looking into it.

Typo: FarmLogs is only tree years old

A very punny one, though :)

For anyone who doesn't know FarmLogs has their HQ in Ann Arbor aka 'tree town'

I'd say just as many tree's here as any college town really.

PS hi chris.

Yea, Always related Ann Arbor with U of M, not trees :}

I never heard it called tree town. The wayback machine page for the Wikipedia citation only says "Tree City USA", which is a thing the Arbor Foundation does for lots of towns.

So either the citation is out of step with the original page or a goofball put that in Wikipedia.

I lived there for 4 years and was aware of "tree town" but never heard anyone call it that. I know it as A2.

I think it has to do with the name 'Arbor' [1] than the town itself, which has a fair amount of trees in the older neighborhoods but isn't excessively forested.

[1]: http://www.oxforddictionaries.com/us/definition/american_eng...

Cool to see tech creeping its way into agriculture.

My in-laws run a vineyard & winery. For many years, I was convinced I would build a sensor-net and on-premises app with some automation capability (toggling servos for irrigation, "fertigation" type system, tanks, timers, etc). Never really made it out of idea stage as my priorities took a different turn, but glad to see some other folks tackling the problems.

Seems like the super low cost of sensors now makes this really attractive (heck, a small farm of couple acres could be wired up to a handful of strategically placed rasberyPi's in a weatherproof case).

This is already reality for decent sized ag setups - measuring things like soil moisture, pH levels etc and reporting back.

This is interesting because there must be a small window between all this data just lying around (cf NSA) and most sectors having the data transformed and pipelined into their decision making.

Sam Altman is right - this is the sort of thing that two software devs can pull off - but like everything the door must be closing.

I'm tempted to make a spreadsheet of all the novel data sources I can find (ONS, satellite, travel, weather etc) and plot against SEC codes - I am pretty sure that will allow me to procrastinate long enough for the enthusiasm to wear off.

> And importantly, they started out doing something that any two programmers (with domain expertise in their market) could have done.

I somehow remember the original text to be

> And importantly, they started out doing something that any two programmers could have done.

There is nothing on the wayback machine either. Am I misremembering the original text?

I remember it without the parentheses, too.

Do these guys have an open data for academic/research purposes?

I remember working out of the same office as these guys back when they were just getting started, and their success and impact has been truly amazing. Recognition well deserved.

Does FarmLogs work for aquaculture?

There's similar project that's focused on cattle: farmeron.com

Is there similar data and service available for European farmers?

Take a look at https://www.365farmnet.com/

Reply on news.ycombinator.com