Hide table of contents

This anonymous essay was submitted to Open Philanthropy's Cause Exploration Prizes contest and published with the author's permission.

If you're seeing this in summer 2022, we'll be posting many submissions in a short period. If you want to stop seeing them so often, apply a filter for the appropriate tag!


  • Movable virtual fencing systems use audio cues with a GPS and data enabled electric shock collar to keep livestock within boundaries and herd them as desired, with many benefits to farmers.
  • Animal welfare organisations are opposed to aversive collars of any kind, but manufacturers claim their products are better for the animals. Currently it is not clear who is right, and it may depend greatly on the design and usage of the systems.
  • This is a new industry with only a few start-ups, but appears to have the potential to rapidly expand and impact the welfare of hundreds of millions of animals.
  • There is potential to fund research at this early stage which could shape the development of these systems, and shape regulation of the industry before it becomes entrenched.
  • Lowering costs of keeping livestock on pasture also has the potential to improve welfare by reducing the number of animals kept in feedlots.
  • Cattle are the initial market, but other pasture animals may be close behind. With lower costs and minimisation perhaps we can even have tiny collars on free range chickens, have free range pigs, or even collar wild animals to keep them safe and monitor their welfare?

Importance of solving the problem

What are movable virtual fencing systems?

Movable virtual fence systems for livestock are systems for containing and herding large farm animals (typically dairy cows) via GPS and data enabled collars capable of delivering electric shocks. In contrast to virtual fencing systems with fixed boundaries, the farmer can use the data connection to move boundaries in real time.

The collars work in a similar way to the common buried wire virtual fence systems used for pet dogs where when the animal approaches the boundary a warning tone is played by the collar which increases in intensity as the animal gets closer to the boundary before finally giving the animal an electric shock. This means that after a learning period the animal should associate the tone with a painful electric shock and so move back from the boundary on hearing the tone without being shocked. The animal can then learn where the boundary is and avoid encountering the warning tone if desired.

With movable boundaries the farmer can remove, create, or (hopefully slowly) move boundaries for the herd or for individual animals. This means that the animals must encounter the warning tones more often as they must experiment to find the new location of the boundaries, or because the farmers are moving the boundaries to ‘herd’ the animals to another location. To make this more predictable for the animals more complex cues can be used. Halter for example appears to use a combination of directional sound and vibration cues to indicate to the cows what direction the boundary is in and encourage them to move in the correct direction (Daniel, 2020) (Stevenson, 2018).

This means the farmers can now do many things that would usually require going to the field from the comfort of their office. Such as: 

  • Move herds between fields
  • Automate movement to and from the milking shed
  • Separate animals from the herd for checking
  • Monitor animal health
  • Monitor pasture utilisation

This provides significant benefits to the farmers, e.g. 

  • Reduced labour
  • Better utilisation of pasture
  • Fewer fences (and so less maintenance)
  • Reduced use of fertilizer
  • Reduced veterinary costs
  • Cheaper compliance with environmental regulation (e.g. protection of waterways)

What is the animal welfare concern?

The obvious concern is that electric shocks are painful and putting collars that cause electric shocks on animals will reduce their quality of life. However, it becomes a bit more unclear as you can make several just so arguments to argue for and against:

  • It’s the same - They should only be shocked during the initial learning phase, and livestock already learn to avoid electric fences by being shocked.
  • It’s worse – Even if they are just getting audio warnings, that warning is associated with pain and so is unpleasant for the animal in a way that seeing an electric fence is not.
  • It’s better – Activity and health can be monitored so hurt or diseased individuals will not suffer for as long.
  • It’s worse – Livestock can see electric fences, and dogs can learn where buried boundary cables are – but farmers can shift the boundaries arbitrarily leaving animals with no idea where the current boundary lies leading to stress and uncertainty.
  • It’s better – Herding livestock with bikes and dogs is more unpleasant than with audio cues.
  • It’s worse – Being remote controlled via Pavlovian conditioning by a capricious god takes away free will, reducing welfare in a more… philosophical sense.
  • It’s better – These high-tech platforms have safeguards to ensure that animal welfare is maintained, probably to a higher standard than your average farm, and farmers cannot modify the strength and frequency of the cues in ways which would hurt the animals.
  • It’s worse – The next entrant in this unregulated industry could come up with a popular system that is terrible for the animals but somehow increases production by a few percent, or farmers could create DIY systems with no safeguards.


What is the positive welfare potential?

Alternatively, whether virtual fencing systems are better or worse for pasture cows – cows appear to prefer pasture over the indoors (Crump, 2021) and intensive feedlots can be terrible for cows welfare (Salvin Hannah E., 2020). Virtual fencing systems have the potential to make pasture farming more economically viable in some cases, particularly on marginal land like mountainous regions, forests, and deserts - where cows must be widely distributed and are difficult to herd. 

If pasture farming with virtual fencing can be cheaper than feedlots and other indoor feeding then there is potential to improve welfare as farmers reduce their use, are driven out of business by lower prices, or the worst offenders can be regulated out of existence now there is an alternative option.

Why now?

Trials of movable virtual fences for cattle have been around since the year 2000 (USDA, 2000) (Cecilie Marie Mejdell, 2017), however they are only now being successfully commercialised. Perhaps because they required modern low power GPS electronics, solar panels, and mobile data networks to be mass produced at low cost.

There are four startups I could find which are competing to establish this new market and given the significant benefits to farmers and the relatively low cost compared to other farming equipment rapid expansion is likely. 

  • Halter, Auckland, New Zealand. They list 9 farms on their website and states they are “deploying to new farms weekly”. (halterhq.com)
  • NoFence, Batnfjordsøra, Norway. They list 3 trial farms on their website. (nofence.no)
  • eShepard, Camberwell, Australia. They state they are ‘performing well at test sites across Australia and New Zealand’ but details are more limited as they have recently been acquired.
  • Vence, San-Diego, USA. They appear to only have a single trial farm? (vence.io)

However there has been little independent research into the welfare of the animals so far, and the market is still in its infancy. Therefore, there is potential to fund research into welfare to shape the emerging products and the regulation surrounding it before the industry becomes entrenched.

How big of a problem?

Livestock numbers and market penetration

There are approximately 1.5 billion cattle in the world (Esri StoryMaps team, 2018), but the majority are raised in Brazil and India which do not appear to be initial targets for the virtual fencing manufacturers. That still leaves a potential initial market around 100 million in the USA (Shahbandeh, 2022), 77 million in Europe (Eurostat, 2020), and another 38 million in Australia (PWC, 2011) & New Zealand (Statistics New Zealand, 2019). Also, these systems are expensive, rely on mobile data coverage, and only really fit into farms which already have quite a high ‘tech level’ so are likely to be limited to large farms in wealthy areas. 

So what kind of market penetration can we expect? Robotic milking machines are perhaps most similar in labour saving benefits and ‘high-tech-ness’. They are only used in around 5% of US dairy farms (Mulvany, 2018), but are already used in around 25% of Danish farms (H.W. Barkema, 2015) and make up a large percentage of new milking machine sales. So there is a good chance virtual fencing systems will follow a similar trajectory from only the biggest farms in wealthy areas, to smaller farms in less wealthy areas as the costs drop and the general tech level in farming increases over time.

Regarding feedlots, beef cows typically start their lives on pasture with their mothers but are then often transferred to feedlots to rapidly increase their weight before sale. In 2022 there were approximately 12 million cows ‘on feed’ in the US at any one time (USDA, 2022), other countries have a smaller but expanding percentage of cows on feedlots (Bechetel, 2018). 

There is potential for using collars on other pasture fed animals as well such as sheep (1.2 billion), and goats (1 billion).

Possible size of effect

Hard to say at this stage, but it can’t be too large a positive or negative effect for the current systems or it would be obvious in some of the case studies which already exist. However, this is only for the systems currently being developed and only used in closely monitored trials. And given how many animals are mistreated around the world in the name of profit it is easy to imagine implementations which are much worse, which I think is the main thing we would be seeking to prevent.

Current regulation

As these are new products there does not appear to be any existing regulation specific to them, however in some areas it would be covered by other similar regulations against electric shock devices such as cattle prods and ‘cow trainers’. For example:

  • They are currently illegal in some Australian states, and research exemptions can be granted in others (RSPCA Australia, 2021).
  • Not covered by existing regulation in Norway (Cecilie Marie Mejdell, 2017).
  • Presumably legal in New Zealand since none of the articles mention it.
  • Presumably also legal in the United States, it doesn’t appear to be covered by the Animal Welfare Act but maybe varies state to state?


Who else is working on this problem?

Some of the manufacturers appear to be making some efforts. 

  • The NoFence system has appeared in a number of case studies and reports, some of which have been concerned with welfare (Brunberg, 2013) (Cecilie Marie Mejdell, 2017) (Magnus Fjord Aaser, 2022).
  • The eShepard system has also appeared in some research papers (Dana L. M. Campbell, 2019) (Caroline Lee, 2021).
  • Halter states “The development of Halter’s smart collar has been independently reviewed and approved by the AgResearch Animal Ethics Committee.” They are presumably conducting internal research given the content of their job postings, but I cannot find any published research.
  • Vence.io don’t mention welfare often, and I couldn’t find any published research. In their terms and conditions they say you must abide by their animal welfare policy – but as of now the page is blank… which is not a great sign.

The University of Western Australia is seeking a PhD candidate to investigate virtual fencing systems, one goal of which is to investigate welfare. (Vercoe, 2022)

RSPCA Australia is opposed to all electronically activated devices that deliver shocks to animals, and for virtual fencing suggests regulation, standards, and third-party monitoring to protect animal welfare. (RSPCA Australia, 2021)

What about other potential funders?

I think most animal welfare organisations and regulatory bodies are generally reactionary and so new technologies like virtual fencing may only become a concern once they are become common in their jurisdictions. I also think many animal welfare organisations would find it difficult to fund research which involves placing aversive collars on animals, particularly if a possible outcome of that research was recommending more use of aversive collars.

Manufacturers do not want negative publicity (so want to ensure their systems meet some minimum standards) but are not incentivised to improve animal welfare at the cost of profits. They are also not incentivised to publicise animal welfare concerns about their products, promote competing products, or suggest regulation of their industry.

Perhaps if virtual fencing becomes established then research funded by animal welfare organisations into the real-world usage of the systems (particularly if there any some truly terrible animal welfare cases which capture the public’s imagination) will lead to regulation and industry standards which provide most of the benefits early research could provide.

I think ambiguous technology like this that has the potential to have a small-to-medium negative or positive effect on the welfare of a very large number of farm animals is precisely the area which is typically not covered by other funders and so is worth investigation by Open Philanthropy.

Potential Interventions

Research into welfare and economics

An initial intervention could be funding independent research into the real-world welfare of animals using virtual fencing systems. By collaborating with both animal welfare organisations and manufacturers, trust in the results by both parties could perhaps be maintained while also sharing the costs. Large, real word trials, with each of the three currently available systems could hopefully definitively answer welfare questions about virtual fencing before the industry becomes entrenched.

Some possible research questions to answer:

  • Are virtual fencing systems as they currently exist better or worse for welfare than standard farming practices?
  • Do the current welfare preserving restrictions built into the systems properly handle edge cases and prevent misuse by farmers?
  • Are there any likely ways a future virtual fencing system could be designed or misused to decrease welfare?
  • Are there any improvements to the system which could improve welfare?
  • What regulations and standards are appropriate?

Additionally funding an economic study into the economics of virtual fencing versus feedlots could determine if it is likely to lead to lead a reduction in feedlot cows once the virtual fencing industry is established.

Outcome: movable virtual fencing is worse than standard pasture farming

  • Work with manufacturers to modify their systems to reduce the hit to animal welfare as much as they are willing to.
  • Lobby to regulatory bodies to restrict the use of virtual fencing systems to where normal fencing is not viable.
  • Lobby to regulatory bodies (or work with industry to self-regulate) to apply regulations to the operation of the systems which reduce the negative effect on welfare as much as possible.

Outcome: movable virtual fencing is better than standard pasture farming

  • Work with manufacturers to improve any negative welfare edge cases discovered.
  • Lobby to regulatory bodies to allow wide adoption of virtual fencing and work against any lobbying efforts by organisations that desire the status quo.
  • Lobby to regulatory bodies (or work with industry to self-regulate) to apply regulations to ensure systems are not designed or misused in ways that hurt welfare.
  • Fund pilot programs in new regions and otherwise work to accelerate uptake.

Outcome: wide use of movable virtual fencing would reduce use of feedlots

As above, plus:

  • Fund pilot programs and work in regions where it is most likely to cause switching away from feedlots.
  • Lobby regulatory bodies to apply more welfare restrictions to feedlots now an economically viable alternative exists.

Outcome: movable virtual fencing is similar to standard pasture farming and is not likely to reduce use of feedlots.

  • Do nothing

Examples of possible regulations on virtual fencing

As the virtual fencing products are currently service / software / hardware platforms many regulations could be applied to the platforms themselves, rather than having to police farmers use of them.

  • All animals on farm must have collars to prevent uncollared animals luring collared animals across boundaries.
  • Restrictions on minimum size of boundaries, frequency of boundary movements, speed of boundary movements.
  • Restrictions on separating of animals from herd.
  • Restrictions on strength of shock, failsafes for injured or trapped animals, volume of audio cues.
  • Boundaries cannot be always totally invisible so animals can know where is safe easily most of the time (e.g. lines on the ground, colour coding, natural boundaries).
  • Restrictions to use on suitable animals only (species, age, gender etc…).
  • Minimum requirements for the cues to ensure the animals have the information they need to avoid the boundaries (e.g. directional cues, different tones, vibration)

Other species and wild animal welfare

Initially cattle seem to be the main target of virtual fencing companies, but there is perhaps scope to improve welfare in other large livestock such as sheep, pigs, deer, and reindeer. In future with lower costs and minimization perhaps even small animals such as free-range chickens.

Another interesting possibility is the use of virtual fencing systems to improve wild animal welfare. Collaring wild animals is of course practically difficult, but perhaps for herd animals only a small number of lead animals must be collared or automated collaring systems could be developed. Possibilities include:

  • Keeping animals away from humans, roads, and farmland allowing larger populations without conflict
  • Monitoring health to prevent suffering due to disease and injury
  • Preventing starvation or lack of water by moving herds to new areas


Bechetel, W. (2018, December 7). Drovers. Retrieved from Cattle on Feed Numbers: A Look Around the World: https://www.drovers.com/markets/cattle-feed-numbers-look-around-world

Brunberg, E. B. (2013). The virtual fencing system Nofence - trials 2013. Bioforsk Report.

Caroline Lee, D. L. (2021). A Multi-Disciplinary Approach to Assess the Welfare Impacts of a New Virtual Fencing Technology. Frontiers in Vetinary Science.

Cecilie Marie Mejdell, D. B. (2017). A review on the use of electric devices to modify animal behaviour and the impact on animal welfare. Oslo: Norwegian Scientific Committee for Food and Environment.

Crump, A. J. (2021). Optimism and pasture access in dairy cows. Scientific Reports.

Dana L. M. Campbell, J. M. (2019). Virtual Fencing Is Comparable to Electric Tape Fencing for Cattle Behavior and Welfare. Frontiers in Vetinary Science.

Daniel, M. (2020, September 16). Virtual fences signal shift in cow management. Retrieved from https://www.ruralnewsgroup.co.nz/dairy-news/dairy-management/virtual-fences-signal-shift-in-cow-management

Esri StoryMaps team. (2018). Farm animal planet. Retrieved from https://storymaps.arcgis.com/stories/58ae71f58fd7418294f34c4f841895d8

Eurostat. (2020, September 23). Livestock population in numbers . Retrieved from Eurostat: https://ec.europa.eu/eurostat/web/products-eurostat-news/-/ddn-20200923-1

H.W. Barkema, M. v.-P. (2015). Changes in the dairy industry affecting health and welfare. Journal of Dairy Science.

Halter. (2020, July). Halter Farmers. Retrieved from https://halterhq.com/halter-farmers

Liveris, J. (2021, October 16). Farming without fences the future of agriculture. Retrieved from https://www.abc.net.au/news/2021-10-16/landline-cattle-collar/100543308

Magnus Fjord Aaser, S. K.-E. (2022). Is Virtual Fencing an Effective Way of Enclosing Cattle? Personality, Herd Behaviour and Welfare. Animals.

Mulvany, L. (2018). More U.S. Milk Will Be Coming From Robots. Retrieved from https://www.vnews.com/More-U-S-milk-will-be-coming-from-robots-15278487

Pavlov’s cows: Is this remote-control cow system creepy, or the future of farming? (2018, June 28). Retrieved from The Spinoff: https://thespinoff.co.nz/business/28-06-2018/pavlovs-cows-is-this-remote-control-cow-system-creepy-or-the-future-of-farming

PWC. (2011). The Australian Beef Industry. 

RSPCA Australia. (2021, December 21). What is virtual fencing (or virtual herding) and does it impact animal welfare? Retrieved from RSPCA: https://kb.rspca.org.au/knowledge-base/what-is-virtual-fencing-or-virtual-herding-and-does-it-impact-animal-welfare/

Salvin Hannah E., L. A. (2020). Welfare of beef cattle in Australian feedlots: a review of the risks and measures. Animal Production Science. Retrieved from https://kb.rspca.org.au/knowledge-base/what-are-the-animal-welfare-issues-with-feedlots-for-cattle/

Shahbandeh, M. (2022, Jul 18). Number of cattle worldwide from 2012 to 2022 (in million head). Retrieved from Statista: https://www.statista.com/statistics/263979/global-cattle-population-since-1990/

Statistics New Zealand. (2019). Retrieved from Stats NZ: https://www.stats.govt.nz/indicators/livestock-numbers

Stevenson, R. (2018, June 28). Pavlov’s cows: Is this remote-control cow system creepy, or the future of farming? Retrieved from The Spinoff: https://thespinoff.co.nz/business/28-06-2018/pavlovs-cows-is-this-remote-control-cow-system-creepy-or-the-future-of-farming

USDA. (2000). The Cyber Cow Whisperer and His Virtual Fence. Agricultural Research.

USDA. (2022, January). National Agricultural Statistics Service. Retrieved from https://www.nass.usda.gov/

Vercoe, P. (2022). Implications of virtual herding for cattle management. Retrieved from https://www.uwa.edu.au/Projects/Implications-of-virtual-herding-for-cattle-management





More posts like this

No comments on this post yet.
Be the first to respond.
Curated and popular this week
Relevant opportunities