The risks of knowing from a distance: remote sensing, social-ecological relationships and ecosystem restoration

Rose Pritchard, Charis Enns, Tim Foster and Laura A. Sauls

Remote sensing data are valuable inputs to decision-making in ecosystem restoration. But as it becomes easier to collect detailed data from afar, is this lengthening other distances too – between restoration advocates and the lands they seek to restore, and between local people and centres of decision-making power?

A savanna restoration project in KwaZulu-Natal, South Africa. Photo: Rose Pritchard

Ecosystem restoration is having a moment in the political sun. The United Nations have designated 2021 to 2030 the Decade on Ecosystem Restoration. And restoration emerged as a priority in the recent COP26, with world leaders pledging to ‘Conserve forests and other terrestrial ecosystems and accelerate their restoration’ as part of the Glasgow Leaders’ Declaration on Forests and Land Use.

Remote sensing data from satellite or near-earth sensors can contribute to many aspects of ecosystem restoration. These technologies make it possible to collect information on land cover change and other key parameters at scales that would be impossible using ground-based survey methods alone. These data may be used to help prioritise restoration locations, plan restoration activities, and monitor the impacts of restoration efforts.

But data and the information products derived from earth observation are political objects. Data in and of themselves are neither good nor bad. The nature and use of data are shaped by people, each of whom will have different values and interests. Thus, remote sensing data can serve emancipatory, empowering purposes in landscapes undergoing restoration; for example, when local people use these data to advocate for their interests. Or the same kind of data could be used to reinforce asymmetrical power relationships and drive exclusionary, inequitable restoration approaches. This latter concern is particularly intense in contexts where restoration activities have already caused harm to marginalised people, such as by restricting access of poorer households to land or leading to displacement of communities.

In our seedcorn project, funded by the Centre for Digital Trust and Society at the University of Manchester, we aim to lay the groundwork needed to address three critical questions on the use and impacts of remote sensing data in restoration, focusing specifically on satellite remote sensing data. First, how are satellite remote sensing data being used by restoration practitioners? Research papers on applications of remote sensing data are multiplying rapidly, from global-scale analyses down to landscape-level studies. But how are these data actually shaping restoration interventions? And how, if at all, is knowledge derived from remote sensing data being combined with other forms of knowledge (such as local ecological knowledge) in the day-to-day practices of ecosystem restoration?

Second, how does use of remote sensing data alter the relationships between restoration practitioners and the lands they seek to restore? That distant imagery can change the way we think about the world is demonstrated by the example of the Earthrise photo, which made clear the fragility of our place in the galaxy and led to a step-change in western environmentalism. However, some things, such as local environmental values, can never be represented through remote sensing imagery. Additionally, satellite remote sensing data are imperfect, and even the highest resolution data still contain uncertainties that can lead to misinterpretation of landscapes. We seek to understand the consequences of the fact that some restoration practitioners and researchers are relating to their target landscapes increasingly – or in some cases exclusively – from a distance.

And third, what does use of remote sensing data mean for relationships of trust and power, both within and beyond landscapes undergoing restoration? A fear raised for conservation more generally is that digital data sources and ‘conservation by algorithm’ will exclude local people from decision-making processes. There are also concerns around privacy, as more aspects of peoples’ livelihoods are being observed without their knowledge or consent. But so far, few studies have drawn out these social and procedural implications of satellite remote sensing data use, in general but particularly in ecosystem restoration.

The potential of remote sensing data to help address the challenge of environmental degradation is undeniable, as demonstrated by the proliferation of increasingly detailed and sophisticated remote sensing applications and tools for environmental purposes over the last two decades. However, the use of such data is not without risks or challenges. Questioning the consequences of how satellite remote sensing data are used, by whom, and with what impacts is an important step towards ensuring that these data support just and lasting approaches to ecosystem restoration – not approaches associated with exclusion and harm.

If you are interested in this project and would like to be part of these conversations, please contact


Distribution of Income from Motorcycle-Based Gig Work in Indonesia

When a consumer pays for motorcycle-based gig work, where does the money go?

Following the approach of an earlier, similar post on car ride-hailing,  and again using data gathered by the Fairwork Indonesia team in Jakarta, we can break this down using the generic model shown below:

a. Amount paid by customer: the service payment plus a platform fee (sometimes called an order or service or transaction processing fee) plus – sometimes – a tip.

b. Amount paid to platform: platforms typically take a commission (a set percentage of the customer service payment, usually between 10-25%) and often also charge a platform fee.

c. Amount paid to worker: all of the tip and the service payment minus the platform’s commission.  In some instances – at the end of a shift or at the end of a week – the worker might also get a bonus payment from the platform e.g. for completing a certain number of tasks or being available for work consistently and/or at particular times.  There may also be other criteria that impact access to bonus payments such as low order cancellation rates or high customer feedback ratings.  Bonuses are paid to the worker from the platform’s share which is taken from the platform’s commission; sometimes also from the platform fee; and in some instances more than this (in other words, in these cases, the worker earns more than the amount paid by the customer due to an additional subsidy taken by the platform from investment or other sources of capital).

The two charts below show the distribution of customer payments for two motorcycle-based gig work platforms (which were charging a 20% gross commission on the customer service payment plus a fee).  Figure 1 presents data for riders who own their own motorcycle (the majority of riders in our sample).  Figure 2 presents data for riders who finance their vehicle through loan repayments or (less frequently) rental.

We can draw a number of conclusions:

i. Shares of the Pie: the worker’s true net income (i.e. after work-related costs have been taken into account) is a significant share – around two-thirds – of the total payment made by the customer.  Aside from the net income earned by the worker, the great majority of the customer payment is captured by large private businesses; typically multinationals – the platform, fuel companies, vehicle finance houses, telecom providers.  A significant chunk of vehicle servicing and maintenance costs even goes this way via parts, oil, tyres, etc.

ii. Fuel Costs: fuel makes up a very significant proportion of costs: around 80% of costs for bike owners; about half of costs for those who finance their motorcycle.  It is therefore not surprising that the price of fuel is always at the forefront of workers’ minds: a relatively small rise can cause quite a significant reduction in their net income.

iii. Financing vs. Owning: as expected, the net income of those who finance their vehicle is a lower proportion of customer payment than that of vehicle owners.  In absolute terms, these two groups take home about the same net income (non-owners’ net income was about 5% lower).  It’s not completely clear how this happens but one contributing factor is that workers who finance their bikes work longer hours in order to help towards earning the extra to cover their repayments: an average 78-hour week compared to a 66-hour week for those who owned their bikes.

iv. Bonuses and Platform Subsidies: as noted below, the figures here are calculated on the basis of 23.5% of rider income deriving from platform bonus payments.  The platform gross commission plus fee represent just over 32% of the customer payment; yet the platform’s net earning is 5% or 6% only.  In other words, and absent unknown factors, the platform is on average paying substantially more than its entire commission to workers.

On this basis, one can calculate the tipping point at which platforms earn nothing and are having to subsidise worker income from investment or other sources of capital.  As illustrated in Figure 3, for this instance, this will happen when worker bonuses make up more than 30% of their income.  Yet one can find examples in Indonesia where the effect of bonuses is to more than double workers’ basic pay (i.e. bonuses make up more than 50% of worker income).  In such circumstances platforms must be significantly subsidising gig work from capital. If this is widespread, it may help to explain why so many gig work platforms report operating losses.

Network effects – the greater value of a platform to users as more users participate – would predict the emergence of monopoly (single seller of services to customers) and monopsony (single buyer of services from workers).  Yet this has not happened in most gig economy markets – including those of Indonesia – which, instead, are oligopolies/oligopsonies, meaning there is competition between platforms for both customers and workers.  It is that competition which in part motivates the payment of bonuses to workers.


– Although insurance is shown as 0%, there are small payments against this item by some workers; just that they are so negligible a component that they rounded down to zero percent.

– The average figures we have included are that 25% of rider income is made up from tips and bonuses, of which tips make up 1.5%.  This must be seen as a very rough-and-ready average because platforms’ bonus payment schemes are continuously changing; their availability typically varies between workers (e.g. with tiered systems such that the highest bonus payments are only accessible by workers who meet particular criteria on workload, availability, cancellation rates, customer ratings, etc.); and workers’ ability to meet the targets necessary for bonus payment varies from day to day.  Bonuses are typically also only achievable for those working very long shifts: some of our sample were working 15- and in a couple of instances 18-hour days.

– The figures here do not take into account any customer-side promotions that platforms occasionally run; the assumption being that these may not alter the share of rider income.

– Fairwork data from South Africa showed riders’ net income to be 55% of the total customer payment, but this did not separately account for bonuses, which will increase the percentage.  Overall, distribution of income will vary between platforms and locations so the figures above should be seen as illustrative rather than universal.

Post by Richard Heeks, Treviliana Putri, Paska Darmawan, Amri Asmara, Nabiyla Risfa, Amelinda Kusumaningtyas & Ruth Simanjuntak.