77. To Uber or not to Uber?

Technology Trends for Food/AgTech leaders

“Software is Feeding the World'' is a weekly newsletter for Food/AgTech leaders, who want to stay abreast of technology trends.

This week’s edition includes the following topics,

  • Analysis: Tracking emissions - it’s complicated and so don’t be reductive.

  • Technology trends: Insure and BNPL in agriculture in the US and India, positional accuracy using OpenARC and Trimble, sustainability matrix.

  • Research review: Improving search space in genomic research by looking at evolutionary patterns across different species, challenges with positional accuracy, and a comprehensive measurement of agricultural sustainability at the country level.

  • Tidbits from the world of Food/AgTech: Tech, M&A, Sustainability.

  • Read, Listen, & Watch: Agriculture research and social benefits, Tim Hammerich’s podcast on drone spraying, and a vintage 1969 Norman Borlaug documentary.


Tracking emissions, one Uber ride, and one bread at a time

To Uber or not to Uber?

A recent study published in the journal Environmental Science & Technology, shows when a single person uses a ride hailing app, all of society pays!!

A car trip made via a ride-hailing app such as Uber or Lyft has external costs that are 30-35% higher than a comparable trip made via a personal vehicle, according to a new study. The ride-hailing app trip has hidden costs to society to the tune of 32-37 cents more, on average, from the impact of pollution and traffic.

This result is counter-intuitive, as app based trips reduce traditional air pollutants (for example sulphur dioxide, nitrogen oxides, and tiny particles) compared to personal vehicle trips, as vehicles used by app drivers are newer, and cars produce more pollution and run less efficiently after sitting for 12 hours or more.

The benefits are canceled by deadheading, when an app-based driver travels from one drop-off to the next pickup, or is simply driving around waiting for the next ride.

The impact of deadheading amounts to a 20% increase in fuel consumption and greenhouse gas emissions and a 60% increase in external costs from traffic (congestion, crashes, and noise) for app-based trips compared to personal vehicle trips.

The researchers suggest taxes and policy approaches to hold down the external costs of app-based travel by encouraging ride-pooling, encouraging app-based travel to destinations where parking is in short supply, and discouraging routes well served by public transit. The issues highlighted by this research are accounted for in the decision making process of drivers and riders, and so doing policy interventions seems like an overreach in this case.

The research focused on a narrow question of comparing GHG emissions of personal rides vs. app-sharing rides. Even though this is not related to agriculture, this research highlights the complications of thinking about sustainability through a very narrow lens. There are significant positive externalities related to ride sharing. Ride-sharing has created a source of flexible income for thousands of people (though there are complications on employee classification). Ride sharing has led to increased driver safety, and reduction in accidents.

Ridesharing has decreased US alcohol-related traffic fatalities by 6.1% and reduced total US traffic fatalities by 4.0%.

The ride-sharing example builds a narrative of pushing the decision whether one should drive their own car, or call an Uber, to the consumer. It seems inappropriate, as it ignores all the positive externalities associated with ride-sharing.

I am writing about this non-agriculture example here to show the issues in food and agriculture systems are even more complicated than the Uber example above. We cannot be influenced by reductive thinking, or big bold headlines. We cannot look at one dimension only, as food and agriculture is a multi-dimensional problem. (more on this in the next section)

Reference: Air Pollution, Greenhouse Gas, and Traffic Externality Benefits and Costs of Shifting Private Vehicle Travel to Ridesourcing Services, Jacob W. Ward, Jeremy J. Michalek, and Constantine Samaras, Environmental Science & Technology 2021 55 (19), 13174-13185, DOI: 10.1021/acs.est.1c01641

Carbon neutral bread?

A 2020 McKinsey report showed wheat is by far the least carbon intensive staple food source already. Nuts and seeds are 7 times more GHG intensive than wheat, rice 13 times, milk 14 times and beef 231 times.

So can bread become carbon neutral?

There are three segments of the supply chain to be considered. Grain production and handling, flour milling and baking. Nearly 2/3rd of the GHG emissions associated with bread production are released on the farm. CPG companies, and their suppliers (for example, growers, and millers) have been stepping up their cooperation.

The push to go carbon neutral is being driven by consumer preferences, and sustainability goals. Companies like General Mills, and off-takers like ADM and Cargill are committed to increase the volume of wheat the companies buy is cultivated using regenerative agriculture practices to reduce GHG emissions. These programs include paying growers to plant preferred wheat varieties, incentives for cover cropping, knowledge and tools, and training programs, though most of them are not for wheat. On the milling side, ADM’s overall US flour milling operations for wheat have achieved carbon neutrality.

There are many challenges for bread to become carbon neutral, especially in the United States.

Wheat acreage planted using regenerative techniques remains a tiny fraction of what is necessary to mill basic bread flour in the United States. Regenerative agriculture is less likely to be embraced in certain areas, including large swaths of the spring wheat states, where wheat is rotated with crops like sugar beets requiring cultivation in a way which disrupts the carbon in the soil.

We can have our bread, and eat it too without guilt.

Prime Future Newsletter

Prime Future is a weekly newsletter for innovators in livestock, meat and dairy. Janette Barnard explores the trends that are shaping the future of animal agriculture and the technology enabling that future.

Janette writes an extremely thought provoking newsletter every week and it's free!

I did a podcast with Janette on Friday about her writing process. I will publish the link to the podcast in next week’s newsletter.

Technology trends

Insuretech, and BNPL in agriculture

Stable, an insurtech aiming to help minimize a businesses’ risk due to volatile commodity prices, today announced it has raised $46.5 million in a Series A round of funding led by Greycroft.

Only 8% of commodities are available to trade on mercantile exchanges (for example CME), which makes purchasing risk management products such as futures or options contracts difficult without enormous basis risk.

Stable hopes to change it. The company’s parametric platform tracks 5,000 3rd party indices to purchase a policy to protect against price volatility. Contract buyers can select one or more of these indices to customer their contract. The use of parametric insurance makes the process much easier, and the closer the index is correlated to the actual risk, it accurately reflects the client’s real risk.

Use of an index makes the claims process completely automated. This is a financial product driven by huge amounts of data and technology to provide relatively simple protection products for contract buyers. The automation and data driven approach keeps the cost of service, management, and claims process much lower, similar to investing in passive index funds for personal investing, vs. having an actively managed portfolio.

On the other side of the world, AgTech startup Unnati has introduced buy-now-pay-later options for farmers. Financing for operational expenses related to input buying is a huge challenge for farmers in India. Unnati was already testing this with retailers and traders, and plans to extend it to farmers. 

Farm owners can purchase inputs such as seeds, pesticides, and equipment accessories from Unnati-registered dealers and pay for them in equated monthly installments (EMI). Farmers can also book their seeds and other inputs in advance through the platform or dealers. The interest subsidy can also be provided by the agri-input brands purchased by the farmer

Unnati has a network of 16,000 retailers and traders through which farmers who have registered on its platform can purchase inputs and sell their produce, representing 275,000 farmers. Unnati uses a variety of parameters to assess the credit worthiness of farmers, including land holding, and their transaction information on the Unnati platform. This is similar to Amazon Business providing working capital credit to sellers on their platform, based on their transactions and growth on the Amazon platform.

Positional accuracy

“Software is Feeding the World’ talks a lot about automation, autonomy, machine learning on the edge, plant level understanding, and use of precision tools.

Most of the applications, especially autonomous vehicle systems, require positional accuracy better than 10 cm. The positional accuracy requirements are definitely driven by user case, level of manual intervention, and economic costs and benefits of the available resolution.

For example, it is common to use a 10m x 10m resolution satellite image to identify problem areas in a large field, estimate biomass, and do management by exception. Satellite imagery is inexpensive, with many tools available to process and analyze the data, and it acts as a good first pass, especially for large areas.

Real-Time Kinematic (RTK) positioning based on at least two GPS receivers—a base receiver and one or more rover receivers. The base receiver takes measurements from satellites in view and then broadcasts them, together with its location, to the rover receiver(s). The rover receiver also collects measurements to the satellites in view and processes them with the base station data. The rover then estimates its location relative to the base.

If only GNSS is available, (Global Navigation Satellite System) it can provide only meter level accuracy. A navigation system includes an RTK positioning engine and GNSS corrections delivered from a ground-based network of secure base-stations will reliably improve position accuracy to centimeter level accuracy. Platforms like OpenARC’s positioning platforms provide tools for developers.

The streamlined OpenARC positioning platform combines Point One’s Polaris GNSS correction service with ACEINNA’s OpenRTK330 hardware and software solution for developers of autonomous systems in precision agriculture, mapping, surveying and robotics. OpenARC provides high precision and confidence in positioning and localization applications, enabling centimeter level accuracy for challenging tasks such as lane keeping, precision agricultural guidance, and drone landing maneuvers.

No silver bullets for sustainability

If one reads about sustainability within agriculture as a casual reader, it might be easy to think it is all about environmental sustainability. Sustainability is not only about the environment, though it is a part of it. According to the UC Davis SARE program

The goal of sustainable agriculture is to meet society’s food and textile needs in the present without compromising the ability of future generations to meet their own needs.”

The articulation of goals by UC Davis is sweeping in nature. It is powerful as it talks about the time dimension, and it is broad enough at the societal level. But how do we measure if we are moving in the right direction, both at the field level as well as larger areas of interest? (for example, state or country level)

An international team has come up with a quantitative sustainable agriculture matrix to highlight priority areas for different countries to address. The matrix includes three major categories:

  1. Economic

  2. Environmental

  3. Social

The goal of the matrix is to make it easier to measure, track, and provide an accountability framework for policy goals, and actual outcomes on the ground (and underneath.)

The matrix breaks down the three categories, into multiple constituent factors, and provides measurements for sustainable zones for each measure.

Graphical View of the Matrix

Overall, most countries have made significant improvement in their socioeconomic indicators but have shown varying levels of deterioration in their environmental indicators.

The current assessment by country and the trend over the last 26 years shows big agriculture countries like the US, India, and China showing improvements along all the 3 directions (at a macro level), whereas countries like Brazil, Ukraine, and the United Kingdom are showing a decline on the environmental front.  The challenges are different in various countries. For example, due to a heavy reliance on corn in the US, one sees the problem of N surplus in the US, whereas Australia has serious issues with P surplus.

The problem of water consumption is becoming worse in countries like India (no surprise), and China, while land use change is a serious consideration in Brazil due to the Amazon deforestation, but the problem seems to be becoming less acute.

The performance of environmental indicators varies among countries mainly due to the differences in their natural resources, agricultural practices, and development stages. Environmental concerns are especially acute in rapidly developing middle-income countries.

The results reveal priority areas for improvement by each country and show that the trade-offs and synergies among indicators often differ.

Figure 4. The 1991–2016 trajectory of SAM indicators for a subset of countries

For example, environmental factors are a huge consideration in a country like India, and at the same time, tools and policies to improve the socio-economic conditions will drive sustainability significantly. Policies and programs to drive better access to finance, improvements in the supply chain infrastructure, land rights, and access to better inputs, technology, and knowhow will have a much bigger impact in developing countries, compared to developed countries.

We cannot be reductive when it comes to sustainability, as it is a complex problem. Right now carbon markets are all the rage within agriculture. There is a risk to equate carbon sequestration with sustainability, when there are many layers within the sustainability matrix.

There are no silver bullets.

Research paper

Evolutionarily informed machine learning enhances the power of predictive gene-to-phenotype relationships

Inferring phenotypic outcomes from genomic features is both a promise and challenge for systems biology. Using gene expression data to predict phenotypic outcomes, and functionally validating the genes with predictive powers are challenges.

Research published in Nature Communications (Sept 24, 2021) highlights how focusing on genes whose expression patterns are evolutionarily conserved across species, can improve the ability to learn and predict “genes of importance” to growth for staple crops.

At a high level, the approach is about reducing the search space by looking for patterns conserved across species. Researchers have to deal with a huge volume of genomic data. Any approach which optimizes the pool of genes one has to look through to find genes with desirable characteristics is desirable. The approach in this research study looked at Nitrogen Use Efficiency (NUE) in Arabidopsis, a small flowering plant widely used as a model organism in plant biology, and varieties of corn.

This reduces the inputs to the ML/AI models. (known as feature selection in machine learning - feature selection is deciding which attributes have a meaningful impact on the outcomes you are trying to model, as part of the exercise of building machine learning models). This enhances the predictive power of a gene expression-to-trait analysis.

Step 1 feature selection: Phenotypic and transcriptomic data of N-responses were generated from Arabidopsis (lab-grown) and maize (field-grown) under low-N vs. high-N conditions. The expression levels of N-response differentially expressed genes (N-DEGs) conserved in both species were identified via “leave-out-one” approach (Fig. 4) and used as gene features in the machine learning methods in Step 2. This biologically principled approach to reduce the feature dimensions ultimately improved the model performance (Table 1). Step 2 feature importance: We ranked the genes based on (i) the XGBoost-derived feature importance score (left) and (ii) the TF connectivity in a GENIE3 regulatory network (right) constructed from the N-response TFs (Step 1) as regulators and the XGBoost important features as targets. Step 3 feature validation: We validated the role of NUE for eight TFs in planta using Arabidopsis and maize loss-of-function mutants.

The researchers conducted experiments that validated eight master transcription factors as genes of importance to nitrogen use efficiency. They showed that altered gene expression in Arabidopsis or corn could increase plant growth in low nitrogen soils.

The results validate the use of evolutionary conserved but differentially expressed genes improve the performance of ML models to predict NUE.

Reference: “Evolutionarily informed machine learning enhances the power of predictive gene-to-phenotype relationships” 24 September 2021, Nature Communications. DOI: 10.1038/s41467-021-25893-w


Tech, Investments, M&A

Software eating the world? General Motors CEO Mary Barra says software services revenue will be equal or greater than vehicle sales in 10 years. Will the same be true for Ag majors in 10 years? 20 years? Never?

Q3 AgTech venture capital investment and exit round up. 20 digital ag startups raised capital last year, with investors focused on digital agronomy, farm management, & yield prediction software.

Acceleration in pace of genome mapping in crops due to technology advances in gene sequencing and genome assembly. If scientists know what is causing the difference in the phenotype, then they can much more precisely try to create a novel variation more meaningful to the farmer, such as disease resistance or yield improvement.

A functional granulate to transport biologicals.” Anuvia and Novozymes partner to expand delivery of biological advantages.

Pivots go from being rabbits to hares. Valley Irrigation cuts the time it takes for a center pivot to make a revolution by half, from 8 hours to 4 hours.

One more ransomware attack in agriculture. This will become common enough to not be news anymore.


The International Potato Center predicts a 32% drop in harvests of potatoes and sweet potatoes by 2060 due to climate change, while some estimates say coffee growers will lose half of adapted lands before 2050. Rice, the world's most important staple food crop, contributes massively to global warming by releasing methane as it is cultivated. It is threatened by rising seas that could put too much salt into the water that floods rice paddies.

Will this report make you “boil” over? Purdue economist’s show organically raised corn, soybean, more profitable than conventionally raised crops.

Langemeier, M. "Conventional and Organic Enterprise Net Returns." farmdoc daily (11):140, Department of Agricultural and Consumer Economics, University of Illinois at Urbana-Champaign, October 1, 2021.

Avocados, Colombia, and water use. “The avocado’s rapid expansion in Colombia began in 2014 when farmers exported 1,408 tonnes of Hass avocado. The industry has since exploded, reaching a record high of 544,933 tonnes in 2020. It takes about 283 litres of water to produce one kilogramme of avocados in Chile. it is 4 times what is needed to produce a kilo of oranges and 10 times what’s required for tomatoes.”

Read, listen, and watch

📚 “How to improve social benefits of agriculture research” by Marci Baranksi and Mary Ollenburger. There are many parallels with product management, solving for the right problem, and has a heavy dose of constructive criticism of various organizations like the Gates Foundation etc.

🎧 Drones for precision spraying with Daniel McCann of Precision AI.

“When you can make a per-plant level decision it's a complete game-changer...so there's all sorts of new types of completely outside-of-the-box approaches that I think are going to become the de facto standard.”

Another good episode of Future of Agriculture and Precision AI. The conversation highlights the challenges with vision based spraying services.

📺 “Norman Borlaug Man of the Year 1969” A documentary from National Film Archives from 1969.

So, what do you think?

💗 If you like “Software is feeding the world”, please hit the “Like” button at the bottom or share with a friend.

🙏 If you don’t mind answering 3 questions anonymously (2 are optional), I would love to get your feedback.

About me

My name is Rhishi Pethe. I lead the product management team at Project Mineral (focused on sustainable agriculture). The views expressed in this newsletter are my personal opinions.