For the first time, scientists have succeeded in artificially mimicking the ion signalling of heart muscle cells. To

UC Berkeley physicists have introduced phase contrast to the electron microscope, allowing scientists to see much smaller molecules

Norway Pulls AI From Grade School Norway is placing tight limits on generative AI across its schools. Prime

A gene therapy designed to replace a missing brain protein restored normal brain activity and improved behavior in

SpaceX’s Trillion-Dollar Debut Elon Musk now holds a title no one has held before. SpaceX went public on

Voting Machine Report on Hold White House officials have held back a federal report on voting machine weaknesses

Android Developer Verification Timeline Google built Android as the open alternative to Apple almost twenty years ago. Now

Beats Studio Buds Eavesdropping Patch Apple has pushed a firmware update to its Beats Studio Buds that closes

Robotaxis Recalled From Highways Waymo has recalled nearly 4,000 robotaxis and pulled them off highways while it works

A Public Stake in AI Bernie Sanders wants to move trillions of dollars from the biggest AI companies

Researchers have identified previously unknown materials, including a new form of a widely studied clean-energy material, by carefully

The popular framing of algorithmic recommendation, repeated across most contemporary writing on consumer technology, is that personalised suggestions surface preferences the consumer already has. The algorithm watches what someone does, identifies a pattern, and offers more of what the pattern predicts they will like. On this account, the algorithm is a useful but essentially neutral tool, one that accelerates discovery without changing what the consumer would have chosen if left to themselves.

The peer-reviewed experimental evidence accumulated over the past twenty years has substantially complicated this picture.

What the controlled studies have repeatedly shown is that algorithmic recommendations do not merely reflect existing preferences. They actively shape them. The brain treats the recommendation itself as a piece of information, an informational anchor, that influences subsequent judgment regardless of whether the underlying algorithm is accurate or not. The same effect persists when the recommendations are deliberately randomised, when they are intentionally incorrect, and when the consumer has been explicitly told that the recommendations may be unreliable. The mechanism is not driven by the algorithm being right. The mechanism is driven by the algorithm being present.

The Adomavicius experiments

The foundational peer-reviewed study of this effect was published in 2018 by Gediminas Adomavicius at the University of Minnesota, with Jesse Bockstedt at the University of Arizona, Shawn Curley at Minnesota, and Jingjing Zhang at Indiana University, in the journal Information Systems Research. The team designed three controlled experiments to test whether personalised recommendation ratings produced by a recommender system would influence the prices that participants would be willing to pay for digital music tracks.

The first experiment showed participants algorithmic recommendation ratings for a series of songs. The ratings were presented as if they had been personalised to each participant by a recommender system trained on their listening history. The ratings were, in fact, randomly assigned. Participants were then asked how much they would be willing to pay for each song. The team found that the randomly assigned ratings substantially altered the prices participants were willing to pay, even when the team controlled for the participants’ own underlying preferences and demographics. Higher random ratings produced higher willingness-to-pay figures. Lower random ratings produced lower figures. The shift was statistically robust across the sample.

The second experiment used actual system-generated recommendations from a real recommender system, but the team intentionally perturbed the recommendations by introducing controlled errors. Participants saw recommendations that were partially based on their preferences but deliberately shifted away from what an unperturbed algorithm would have produced. The team observed substantially the same effect. The participants’ willingness to pay tracked the perturbed recommendations rather than their underlying preferences.

The third experiment reduced preference uncertainty by giving participants substantial time and information to develop firm views about the songs before they were exposed to the recommendation ratings. The hypothesis the team tested was that participants who had already made up their minds about the music would be less susceptible to algorithmic influence. The hypothesis did not hold. Even participants with stable, well-developed preferences showed the same willingness-to-pay shifts in response to randomly assigned algorithmic recommendations.

The anchoring mechanism

The Adomavicius team’s interpretation, supported by the broader peer-reviewed literature on cognitive anchoring effects, was that the recommendation operates as a numerical anchor that the brain uses as a starting point for subsequent judgment. The classic anchoring effect, identified by Daniel Kahneman and Amos Tversky in the 1970s and replicated in thousands of subsequent studies, shows that arbitrary numerical values presented before a judgment systematically shift the judgment toward the anchor, even when the people making the judgment know the anchor is irrelevant.

Algorithmic recommendations function as anchors in the same way. A recommendation of four out of five stars sets a numerical reference point that influences the consumer’s subsequent valuation of the product, regardless of whether the four-star recommendation reflects anything meaningful about the consumer’s actual preferences. The consumer who has been told a song is worth four stars, by what they believe to be an accurate personalised algorithm, will offer a price closer to what a four-star song commands than they would have offered in the absence of the recommendation. The shift occurs whether the four-star rating is generated by an accurate algorithm, an inaccurate algorithm, or a random number generator.

The implication is that the source of the algorithmic influence is not the accuracy of the algorithm. The source is the presence of the algorithm. The consumer is being influenced by a numerical informational anchor that has been generated by a system the consumer trusts to be measuring their preferences. Whether the system is actually doing that is, on the peer-reviewed evidence, largely beside the point.

The Amazon and YouTube specifics

The peer-reviewed experimental findings sit alongside a substantial body of industry data on the actual scale of algorithmic influence in major consumer platforms. Amazon’s recommendation engine, on figures the company has publicly disclosed and that have been independently analysed by McKinsey and Company since 2013, drives approximately 35 per cent of all purchases made on the platform. The figure has held roughly steady for over a decade, despite extensive evolution in the underlying algorithms and substantial growth in Amazon’s product catalogue.

YouTube’s chief product officer Neal Mohan disclosed at the Consumer Electronics Show in January 2018 that approximately 70 per cent of all viewing time on YouTube comes from videos selected by the platform’s recommendation algorithm rather than from videos that users have actively searched for. The figure has been widely confirmed by subsequent industry analysis and remains the standard reference for the role of algorithmic recommendation in video discovery. YouTube now has approximately 2.5 billion monthly users.

The figures across other major platforms follow similar patterns. Pew Research Center has found that approximately 81 per cent of YouTube users say they at least occasionally watch recommended videos, with 15 per cent reporting that they do so regularly. The recommendation algorithms on TikTok and Instagram measure user engagement to the millisecond and adjust the content stream in real time based on which videos a user has lingered on, replayed, or scrolled past quickly. The effect of all this, on the available evidence, is that the majority of the content the average person encounters in the major consumer platforms is selected by algorithms rather than chosen by the person consuming it.

The manipulation framework

A 2019 peer-reviewed article by Daniel Susser at Pennsylvania State University, Beate Roessler at the University of Amsterdam, and Helen Nissenbaum at Cornell Tech, published in the Georgetown Law Technology Review, argued that the systematic deployment of algorithmic influence in digital platforms now meets the philosophical definition of manipulation rather than persuasion. The distinction matters because manipulation, in the philosophical literature, is defined as the deliberate influence of someone’s beliefs, desires, or actions through means that bypass their rational agency. Persuasion, by contrast, addresses someone’s rational agency directly by giving them reasons to change their beliefs or actions.

The Susser team’s argument was that algorithmic recommendations, as currently deployed, systematically bypass the rational agency of consumers. The consumer is not given reasons why a particular product is being recommended. The consumer is not told what factors the algorithm has weighted, or how heavily, or what alternatives have been excluded. The consumer is presented with a numerical or categorical anchor and expected to make a decision in response to it. The cognitive mechanism the recommendation activates, on the Adomavicius experimental evidence, is the anchoring effect, which is precisely the kind of non-rational decision-making bias that the philosophical literature has long identified as the substrate of manipulation.

The Susser argument has not produced consensus. Other scholars have argued that the line between manipulation and persuasion is blurry, that consumers retain meaningful agency to ignore or reject algorithmic recommendations, and that the consumer welfare benefits of algorithmic curation outweigh the autonomy costs. The peer-reviewed dispute is genuinely ongoing. What is not in dispute is that the cognitive mechanisms identified by the Adomavicius team operate in ways the consumer is generally unaware of, and that they produce substantial behavioural effects that the consumer would not necessarily endorse if they were aware of them.

The honest limitations

Several methodological caveats apply to the literature described above.

The Adomavicius experimental work has been conducted primarily in laboratory settings using digital music as the stimulus. The generalisability of the findings to other product categories, other recommendation contexts, and the open-ended environment of actual consumer life is plausible but not perfectly established. The effect sizes observed in the lab may be larger or smaller than the effect sizes that obtain in everyday digital commerce.

The industry figures for Amazon and YouTube are derived from corporate disclosures rather than from independent peer-reviewed measurement. The 35 per cent and 70 per cent figures are widely cited and broadly consistent with the available academic analyses, but they should be treated as informed estimates rather than as precise scientific measurements.

The Susser 2019 manipulation framework is a philosophical argument rather than an empirical finding. The conceptual definition of manipulation, and the application of that definition to algorithmic recommendation systems, remain genuinely contested in the peer-reviewed legal and philosophical literature. The argument is strong but is not the only available position.

The anchoring mechanism identified by the Adomavicius team is one of several cognitive effects that have been proposed to explain algorithmic influence on consumer behaviour. Other proposed mechanisms include social proof effects, decision fatigue and the deferral of judgment to perceived authority, identity-reinforcement effects in which recommendations mirror and amplify the consumer’s existing self-concept, and choice architecture effects in which the structure of the recommendation environment itself shapes which options the consumer considers. The relative contributions of these mechanisms are not yet fully understood.

What it means

Several things follow from the peer-reviewed evidence on algorithmic influence on consumer choice that are worth saying clearly.

The first is that the popular framing of algorithmic recommendation as a neutral discovery tool is not supported by the available experimental evidence. The recommendation is not a neutral observation of what the consumer would have chosen. It is an active intervention in the consumer’s decision-making process, operating through cognitive anchoring mechanisms that are substantially independent of the recommendation’s accuracy. The consumer who receives a recommendation is being measurably influenced by the fact of the recommendation, not just by its content.

The second is that the scale of algorithmic influence in modern consumer platforms is substantial. Approximately 35 per cent of Amazon purchases, approximately 70 per cent of YouTube viewing time, and the majority of content consumption on TikTok, Instagram, and the major streaming platforms now come from algorithm-driven recommendation rather than from user-initiated search. The figure means that, for most consumers in most digital environments, what they encounter is selected by algorithms rather than by them.

The third is that the cognitive mechanism producing the algorithmic influence is one the consumer is generally not aware of, and one the consumer cannot easily resist even when they have been told the recommendation may be unreliable. The Adomavicius team’s third experiment, in which participants with firm pre-existing preferences still showed the influence effect, suggests that informed scepticism about algorithms does not, by itself, neutralise the cognitive mechanisms by which algorithms shape decisions.

The fourth, on the strongest current reading of approximately twenty years of peer-reviewed evidence in behavioural economics, decision psychology, and the philosophy of digital technology, is that the consumer who believes their choices in algorithmically curated environments are predominantly their own is, on the available evidence, mistaken in a measurable way.

The choices are partly theirs.

They are also partly the algorithm’s.

The peer-reviewed evidence has not yet established the precise weighting, but it has established that the second contribution is larger than the popular framing has assumed.

The post Approximately 35 per cent of all purchases made on Amazon are driven by its recommendation engine, approximately 70 per cent of all viewing on YouTube comes from algorithm-driven recommendations, and peer-reviewed evidence has shown that even randomly assigned recommendations measurably alter what consumers are willing to pay for a product appeared first on Space Daily.

A University of Queensland breakthrough could soon be used to upgrade the world’s most prominent gene-editing technology to

Old Passwords, Fresh Victims Two cybersecurity firms say a single hacking crew has quietly broken into tens of

The popular framing of streaming’s environmental cost has, for much of the past five years, treated digital entertainment as a quiet ecological catastrophe. Articles in the New York Post, the CBC, Yahoo, Deutsche Welle, Gizmodo, the BBC, and dozens of other outlets have repeated a version of the same claim, that watching 30 minutes of Netflix is the carbon equivalent of driving four miles, or that the global streaming industry produces emissions comparable to the entire country of France. The framing is intuitive. It maps onto a wider cultural sense that all internet activity has some hidden but enormous environmental cost, and that the cloud is, in fact, a coal-fired power station with a friendlier name.

The framing is also, on the strongest current reading of the peer-reviewed evidence, wrong by approximately a factor of 90.

The actual peer-reviewed figure for the carbon footprint of one hour of streaming video, in 2019 conditions on the global average electricity mix, is approximately 36 grams of carbon dioxide equivalent. The figure is approximately the same as the emissions from boiling a kettle for six minutes, or driving an average passenger car for approximately 150 metres, or running an older incandescent light bulb for approximately two hours. The cumulative impact of streaming at global scale is genuinely significant, but the per-hour impact is modest, and the actual environmental cost falls in a very different place from where the popular framing has located it.

How the popular framing got so wrong

The widely repeated claim that streaming half an hour of Netflix produces 1.6 kilograms of carbon dioxide originated in a July 2019 report by the Shift Project, a French climate think tank. The report estimated that streaming video produced approximately 300 million tonnes of CO2 in 2018, equivalent to the annual emissions of France, and that streaming one hour produced approximately 3.2 kilograms of carbon dioxide. The figure was picked up across global media and became the foundation of the popular framing.

The figure was also based on a bit-to-byte conversion error. In June 2020, the Shift Project published a follow-up article acknowledging that the original 2019 estimate had overstated streaming’s bandwidth requirements by a factor of eight, and revised the 3.2 kg per hour figure downward to approximately 0.4 kg per hour. The correction received substantially less media coverage than the original error, and the 3.2 kg figure has continued to circulate in popular reporting on streaming’s environmental impact for years after its retraction.

The Shift Project’s corrected 0.4 kg per hour estimate is itself approximately ten times higher than the central peer-reviewed estimate from independent researchers. George Kamiya, a digital energy analyst at the International Energy Agency, published a detailed methodological analysis in December 2020 that identified the remaining assumptions in the Shift Project estimate that produced the residual overestimate. The Shift Project had assumed substantially higher data transmission energy intensity than the available evidence supported, had double-counted certain infrastructure costs, and had used outdated electricity grid carbon intensity figures that did not reflect the rapid decarbonisation of major streaming markets since 2015.

The IEA central estimate, drawing on the peer-reviewed literature and updated infrastructure efficiency data, places the actual figure at approximately 36 grams of carbon dioxide per hour of streaming.

The Shehabi peer-reviewed analysis

The foundational peer-reviewed study of streaming’s environmental cost was published in 2014 by Arman Shehabi of Lawrence Berkeley National Laboratory, with Ben Walker and Eric Masanet of Northwestern University, in Environmental Research Letters. The team applied a full life-cycle assessment methodology to streaming video viewing in the United States in 2011, accounting for energy use and greenhouse gas emissions across data centres, network infrastructure, end-user devices, and the embodied emissions of manufacturing and disposing of all of the above.

The Shehabi team’s central estimate was that one hour of US streaming in 2011 produced approximately 420 grams of carbon dioxide equivalent on a full lifecycle basis, and approximately 360 grams when only operational emissions were counted. The figure was substantially higher than the more recent IEA estimate of 36 grams, but the difference reflects both the dramatic improvement in data centre and network efficiency over the past decade and the substantially different electricity grid mix in 2011 compared with 2019.

The Shehabi team’s most important finding, on the peer-reviewed evidence, was not the headline carbon figure but the breakdown of where the emissions actually came from. Their analysis showed that more than half of streaming’s energy use, in 2011 US conditions, came from network data transmission rather than from data centres, and that end-user device energy was the second-largest contributor. The data centres that store and transmit the content, which the popular framing has consistently treated as the central environmental cost of streaming, were actually a relatively modest contributor to the total footprint.

The Shehabi team also noted that streaming video was already more energy-efficient than the alternative the popular framing was implicitly comparing it against, which was the manufacture, transportation, and physical viewing of DVDs. Streaming used less primary energy than DVD viewing in nearly every realistic comparison. The exception was the case in which the consumer walked or biked to a video rental store, which the team acknowledged was the most environmentally favourable form of DVD viewing but also the rarest in practice.

Where the emissions actually come from

The most striking finding in the recent peer-reviewed evidence on streaming’s environmental cost concerns the distribution of the footprint across the system. Netflix’s own ESG reporting, drawing on detailed lifecycle analyses commissioned from the Carbon Trust and independent auditors, has consistently found that approximately 89 per cent of the carbon footprint of watching Netflix comes from end-user devices, not from data centres or network infrastructure.

The breakdown is substantial. Television screens and other viewing devices account for approximately 46 per cent of total streaming emissions. Home networking equipment, including modems, routers, and Wi-Fi devices, accounts for approximately 38 per cent. Data centres, the part of the system that the popular framing has treated as the central environmental cost of streaming, account for approximately 4 per cent. Network infrastructure between the data centre and the viewer accounts for approximately 7 per cent.

The implication is that the environmental cost of streaming is concentrated in objects the viewer already owns, that the viewer would own and operate regardless of whether they streamed video or not, and that the marginal environmental cost of an additional hour of streaming, given that the screen and the router are already on, is genuinely small. The viewer who replaces a 65-inch plasma television with a modern LED screen of the same size reduces the per-hour environmental cost of their streaming by more than the data centres at Netflix could reduce it through any infrastructure upgrade.

The popular framing has been arguing about the wrong part of the system.

The cumulative scale

The per-hour figures, while modest, accumulate at global scale into something substantial. Streaming video now accounts for approximately 80 per cent of all internet traffic. Netflix’s total annual emissions reached approximately 5.17 million tonnes of carbon dioxide equivalent in 2024, according to the company’s published ESG reporting. The global streaming industry, including YouTube, Amazon Prime Video, Disney Plus, and Spotify, produces emissions comparable to the annual output of a mid-sized industrialised country.

The information and communication technology sector as a whole, including streaming, social media, cloud computing, cryptocurrency, and the energy used by devices to access all of the above, now accounts for approximately 1.9 per cent of global greenhouse gas emissions. The IEA projects that global data centre electricity consumption will reach approximately 945 terawatt-hours by 2030, which would be approximately 3 per cent of global electricity consumption.

The 1.9 per cent figure is substantial but is also, on the available comparison data, smaller than several other categories of activity that the popular framing has generally treated as more virtuous. Aviation accounts for approximately 2.5 per cent of global emissions. The fashion industry accounts for approximately 4 to 8 per cent depending on the methodology. Food waste accounts for approximately 8 per cent. The streaming industry’s environmental cost is real but is not, on any defensible reading of the peer-reviewed evidence, larger than the environmental cost of many activities that produce considerably less public concern.

The honest limitations

Several methodological caveats apply to the literature described above.

The first is that the 36 grams per hour figure is a central estimate based on global average electricity mix in 2019. The actual figure for any specific viewer depends substantially on where they live, how their electricity is generated, what device they use to watch, and what video resolution they choose. A viewer in Iceland streaming standard-definition video on a modern laptop produces approximately one tenth the per-hour emissions of a viewer in West Virginia streaming 4K video on an old plasma television. The popular framing’s tendency to quote a single number obscures genuine variation that depends on choices the viewer can actually influence.

The second is that the 89 per cent figure for end-user device contribution reflects current grid mixes and device efficiencies. As data centres expand to meet rapidly growing artificial intelligence workloads, the balance is shifting. Recent IEA projections suggest that data centre electricity demand may rise faster than the corresponding efficiency gains, and the share of streaming’s footprint attributable to data centres may grow over the next decade even as the absolute per-hour figure continues to fall.

The third is that the lifecycle assessment methodology used by the Shehabi team and others depends on assumptions about device lifetimes, replacement rates, and recycling patterns that are difficult to verify and that vary substantially across populations. The peer-reviewed figures are best treated as informed estimates with substantial uncertainty rather than as precise measurements.

The fourth is that several of the most widely cited sources for streaming’s environmental cost, including the Carbon Trust 2021 analysis that produced the 55 grams per hour Europe figure, were commissioned or co-funded by streaming companies themselves. The findings have been broadly consistent with independent peer-reviewed estimates, but the funding source is worth noting.

What it means

Several things follow from the differentiated picture of streaming’s environmental cost that are worth saying clearly.

The first is that the popular framing of streaming as an environmental catastrophe is, on the strongest peer-reviewed evidence, substantially overstated. The 3.2 kilograms per hour figure that has circulated in major media outlets for years is approximately 90 times higher than the actual peer-reviewed estimate. The framing has produced years of guilt-laden coverage of digital entertainment that the underlying science does not support.

The second is that the actual environmental burden of streaming falls on a different part of the system than the popular framing has located it. Data centres, despite their size, dramatic energy consumption, and visible cooling infrastructure, contribute approximately 4 per cent of streaming’s carbon footprint. The screens that viewers watch on contribute approximately 46 per cent. The home networking equipment that connects those screens to the internet contributes approximately 38 per cent. The popular conversation about how to reduce streaming’s environmental impact has consistently focused on the wrong part of the system.

The third is that the cumulative scale of streaming remains substantial. The 1.9 per cent of global emissions from the broader ICT sector is comparable to the airline industry. The 945 terawatt-hour projected data centre electricity consumption by 2030 is approximately 3 per cent of global electricity. Streaming is not an environmental free pass at scale, even though the per-hour cost is modest.

The fourth, on the strongest current reading of approximately ten years of peer-reviewed evidence and several methodological corrections to the popular framing, is that the most effective way for an individual viewer to reduce the environmental cost of their streaming is not to stream less, but to watch on a smaller and more efficient screen.

The popular framing has been telling people to stop watching Netflix.

The peer-reviewed evidence suggests they should instead buy a smaller television.

The post The environmental impact of watching one hour of video on Netflix is approximately 36 grams of carbon dioxide, roughly equivalent to driving a car for 150 metres, in a figure substantially smaller than the popular framing has claimed and one that demonstrates that the actual environmental burden of streaming comes not from the data centres people imagine but from the screens they watch on appeared first on Space Daily.

Safer and more environmentally friendly indoor solar panels could soon help power electronics in homes and offices, thanks

Binance Braces for EU Exit Binance, the world’s largest crypto exchange, is about to lose the right to

Germany’s Push for a Homegrown Cloud A growing share of German companies would accept a weaker cloud service

Android 17 Reaches Pixel Google started pushing Android 17 to Pixel phones today, and the update arrives alongside

Experts Urge U.S. to Lift Export Ban on Anthropic’s Fable and Mythos Models Dozens of prominent cybersecurity practitioners