sure. Mitigating the Jevons paradox requires creative approaches that may include substitution of goods by services and dematerialization, for example, by virtualization. 11 Such changes have the potential to entail a drop in absolute consumption, although so far, most approaches have tended to focus on increasing efficiency, which may or may not result in absolute reductions. 13 However, there is scope for significant change; for example, the energy costs of a virtual meeting that transmits data to a large number of remote participants is tiny compared to the energy cost of a single airplane trip for a single participant. The energy needed for data transmission is decreasing at a fast pace, unlike the energy costs of air travel. Aslan et. al. 2 estimate that data transmission costs decrease by 50% every two years.

Accounting for resource use must be done thoughtfully, with long-term goals in mind, in view of the big picture. There is justification for spending resources during a time of relative abundance to prepare for a future of scarcity. 12 Not all investments need to pay off immediately. There is a place for experimenting when we don’t know for sure if savings will be accrued. But such experimentation should fail fast, and have a plausible hope of saving resources. In this regard, we need to be cognizant of the power of capital markets in deciding what is a success and what is a failure. While markets are very good at optimizing the delivery of the goods and services that they incentivize, they tend not to be organized in such a way that promotes long-term returns or incorporates the costs of the externalities that push limits. Structural changes such as cap-and trade markets, taxes, fees, rationing, and quotas are needed, in concert with technological changes, to address these issues.

Another key approach involves
finding energy savings through disin-
termediation, that is, the process of le-
veraging technology to supplant “mid-
dleman” actors in resource chains. 25
Traditionally, in the absence of infor-
mation technology, such middlemen
provided value and extracted costs
by creating markets and distribution
centers for goods. For example, sys-
tems to directly connect small-scale
worker/producer owned facilities
Such a re-arrangement certainly went
against existing building codes, but
recognized the low cost of alleviating
some resource deprivation in exchange
for neighborhood stability.

A case study such as this can be generative by revealing opportunities for developed and less developed regions to transfer technologies and schemes of sociotechnical organization that present a different set of economic incentives for actors. Being aware of the wide diversity of current and future potential contexts in which humans may find themselves, more than a few of which are characterized by scarcity, may help computing researchers and practitioners design technology that promotes global well-being.

Several other LIMITS-relevant papers have focused on aspects of this principle, including work found in Refs. 4, 14, 29, 30, 40

Reduce energy and material consumption. Sticking to the dominant narrative of growth is riskier than just making a bad guess. It is dangerous because it creates a possibility that we will reach a point at which resources have precipitously dwindled and we may not have enough remaining resources to make the necessary corrections to avert catastrophic outcomes. Therefore, it is important to acknowledge that computing uses energy and material resources. If, as we have argued, these resources are declining, a threshold that LIMITS research should meet is that it is worth the resources it consumes. Put another way, LIMITS research, once applied, should reduce energy expenditures and material consumption. This reduction is difficult to assess, but not something we can sidestep.

More broadly speaking, attempts to limit resource usage in any human system are notoriously challenging. Most of us are well aware of the problems of CO2 emissions, but less aware of more subtle dynamics such as the Jevons paradox, that is, that more efficient technologies often encourage greater use of a resource, reducing or eliminating savings. A more efficient gas engine may reduce fuel consumption by half, but stimulate more than twice as much driving (as well as more cars). A more efficient cryptocurrency mining chip effectively increases electricity consumption through competitive pres-

LIMITS research,
once applied, should
reduce energy
expenditures
and material
consumption.
This reduction is
difficult to assess,
but not something
we can sidestep.