Exploring Privacy and Security as Drivers for Environmental Sustainability in Cloud-Based Office Solutions (Extended Abstract)

Iness Ben Guirat; Jan Tobias Muehlberg; Jason Kayembe

REVIEW 3 major objections 4 minor 18 references

Exploring Privacy and Security as Drivers for Environmental Sustainability in Cloud-Based Office Solutions (Extended Abstract)

T0 review · 3 major / 4 minor · reviewed 2026-08-12 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read This paper argues that a Selenium-and-Green-Metric-Tool framework can rank webmail services by user-side energy consumption, and reports that in its measurements Gmail is slightly more efficient while Proton consumes more energy…

desk verdict An honest, well-scoped workshop paper extending user-side energy measurement to webmail; the concrete Gmail/Proton ranking is plausible but unvalidated against ground-truth power data. read the letter →

arxiv 2411.16340 v2 pith:D4YAFCL4 submitted 2024-11-25 cs.SE cs.CY

Jason Kayembe , Iness Ben Guirat , Jan Tobias Muehlberg This is my paper · ORCID

classification cs.SEcs.CY

keywords privacysecurityenvironmentalsustainabilityenergyconsumptionwebmailad-blockinguser-sidemeasurement

verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

The paper sets out to test whether privacy- and security-focused online services are greener than advertising-financed ones, using webmail as the case study. It builds a measurement framework that automates typical user tasks in a browser and uses container-isolated energy monitoring to attribute device power use to individual services. Applied to Gmail, Outlook, and Proton Mail, the framework finds no significant effect from ad-blocking and yields a preliminary user-side ranking in which Gmail is slightly more energy-efficient and Proton consumes more, especially in memory. The authors are explicit that this does not yet confirm or refute the main hypothesis that privacy-preserving services are greener, because server-side and embodied emissions are estimated from sector averages rather than measured.

What carries the argument

The central object is the six-step user-side assessment pipeline: select comparable services, define functional units, automate them with Selenium, monitor with the Green Metric Tool's container isolation, convert energy to greenhouse-gas emissions via grid intensity, and compare emissions and traffic. The load-bearing mechanism is the Green Metric Tool's claim that enclosing a browser in a lightweight container allows the machine's embedded sensors to attribute energy and data transfer specifically to that container.

What would settle it

Run the same five functional-unit scenarios for Gmail, Outlook, and Proton Mail while measuring total machine power with an external power meter; if the sum of the tool's per-container energy attributions does not track the meter within measurement error, the reported provider differences cannot be trusted.

Watch

Extended reading notes

Core claim

On the paper's own terms, the discovery is that the user-side environmental footprint of an online office service can be decomposed into functional units (idle, login, logout, attachment, reply, delete), automated with Selenium, and monitored per-container with the Green Metric Tool, producing provider-level energy and data-traffic comparisons. In the limited runs, Gmail was slightly more energy-efficient and Proton consumed more energy, particularly in memory, while ad-blocker activation produced no statistically visible change. These results make it possible to say which tested webmail service is greener on the user side, but not yet to settle whether privacy-preserving business models are generally greener, because the dominant server-side and embodied emissions are estimated from sector averages rather than measured.

Load-bearing premise

The entire provider ranking rests on the Green Metric Tool's container isolation correctly assigning the computer's energy and network usage to the single service being tested, and the paper does not validate that attribution against a physical power meter or a known reference load.

Editorial extensions

If this is right

If the framework is correct, anyone can rank webmail and other online office services by user-side energy with automated browser scenarios, without manual instrumentation.
Under these measurements, ad-blocking does not noticeably change user-side energy for webmail, suggesting the main user-side cost is the service's own code and data handling rather than displayed ads.
The functional-unit decomposition lets complex tasks be estimated by aggregating basic scenarios, which could extend to other office tools such as document editors and calendars.
The paper's user-side emissions estimates provide a concrete baseline against which future server-side transparency data could be compared.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

If container attribution is validated against physical power meters, the same framework could make energy scorecards for arbitrary web applications feasible.
The null ad-blocker result may be an artifact of webmail's low ad density; on ad-heavy sites the larger ad-blocker effects reported in earlier studies might resurface.
Because server-side and embodied emissions dominate the life cycle of cloud services, the privacy-greener hypothesis may ultimately be decided by data-center transparency rather than by user-side measurements.
A natural next experiment is to rerun the functional units without wait statements to see whether time savings from ad-blocking translate into energy savings, as the paper itself suggests.

Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Desk Editor's Note

read the letter

Short version: this is an honest, well-scoped workshop paper that combines existing tools into a user-side energy-assessment framework and applies it to Gmail, Outlook, and Proton Mail. The authors do not oversell their main hypothesis; they explicitly say the privacy-green link is unresolved. Their concrete positive claim is that Gmail is slightly more efficient and Proton heavier on memory, and that ad-blockers showed no significant effect.

What I like: the framework decomposition is sensible, the choice of functional units is reasonable, and the limitations section actually engages with the hard problems—server-side opacity, network inelasticity, the ad-blocker's own cost, and the wait-statement distortion. They provide an artifact link, which is more than most. The citation pattern is appropriate: Pesari et al. is the right base, Pearce et al. is the right ad-blocker reference, and Malmodin et al. supplies the external emission factors without any parameter fitting to their own data.

Where it wobbles: the ranking depends entirely on Green Metric Tool's container-scoped energy readings, and there is no validation against a physical power meter or a known-load workload. The authors state the attribution is accurate on the authority of GMT's documentation. With only five runs per functional unit and no reported variance or statistical test, the "slightly more efficient" Gmail result could easily be measurement noise or container overhead rather than a property of the service. The ad-blocker null result is also weakened by the standardized wait statements; the paper acknowledges this. So the paper is a credible first step, not a confirmed finding.

I agree with the stress-test note: the container-attribution question is load-bearing, and a calibration experiment is the missing piece.

Who should read it: people in green software measurement and ICT carbon accounting, especially those thinking about how to attribute user-side energy to specific online services. It deserves referee time at a workshop; as a full journal paper it would need ground-truth calibration, more runs, and versioned artifacts. My call: send it to review, and ask the authors to validate the container attribution before any user-facing ranking is promoted.

Referee Report

3 major / 4 minor

Summary. The paper proposes a multi-level framework for assessing the environmental footprint of cloud-based office services, breaking the assessment into user-side, network/server-side, and embodied/end-of-life emissions. It applies the user-side part of the framework to Gmail, Outlook, and Proton Mail using Selenium-automated functional units and the Green Metric Tool for container-level energy monitoring. The experiments comprise five runs per functional unit and report no meaningful ad-blocker effect, while observing that Gmail is slightly more energy-efficient and Proton Mail consumes more energy, particularly in memory. The paper is explicitly cautious throughout, presenting these results as preliminary and listing limitations for each assessment layer.

Significance. The topic is timely, and the proposed framework is a useful structuring device for comparing the user-side energy behavior of online services. The authors are honest about the preliminary nature of the evidence and make their automation tool open source, which strengthens reproducibility. The use of external per-GB emission factors from Malmodin et al. rather than fitted parameters is a methodological plus. However, the paper's concrete empirical contribution, the ranking of the three webmail providers, is not yet established because it depends on unvalidated container-based energy attribution and on a very small number of runs without statistical reporting.

major comments (3)

[Section 3, Green Metric Tool paragraph] The user-side ranking of Gmail, Outlook, and Proton Mail rests entirely on the Green Metric Tool's container-level energy readings, yet the assertion that container isolation 'accurately evaluate[s] the specific energy consumption' is not validated against a physical power meter, a calibrated workload, or an independent measurement tool. Because the reported provider differences are only 'slightly' different, systematic attribution errors such as container overhead, CPU frequency scaling, or indirect memory-energy estimation could easily overwhelm or invert the ranking. The paper should either add a calibration experiment or explicitly demote the ranking claim to an illustration of the framework until such validation is provided.
[Section 3, experimental results] The experiments consist of five runs per functional unit, but no variance, confidence intervals, or significance tests are reported. The sentence 'no significant differences were found' is therefore not a statistical conclusion; it is an informal observation. The paper should report per-run energy values with dispersion measures and apply an appropriate test, or at minimum show the distributions, for both the ad-blocker comparison and the provider comparison.
[Section 3.1 and Section 2.1, Step 3] Standardizing functional units with wait statements so that all providers run for the same duration makes the ad-blocker comparison uninterpretable as a test of Pearce et al.'s time-savings result, as the paper itself acknowledges. This standardization may also affect the provider ranking, since equivalent user tasks such as reading three emails, replying, and logging out can take different wall-clock time and involve different amounts of work across providers. Re-running without artificial waits, or reporting both wall-clock time and energy per task, is needed before the provider differences can be attributed to service design rather than to the test harness.

minor comments (4)

[Section 3, functional units] The text states that 'Six fundamental functional units were defined' but then enumerates seven items: Idle, Login, Logout, No attachment, Attachment, Reply, and Delete.
[Title and abstract] The paper contains typographical artifacts in the title and header, including 'Environmenta l' and 'Oﬀice'; these should be corrected.
[References [7] and [8]] References [7] and [8] carry the same title and URL; [8] should point to the Energy ID project page if that is intended.
[Section 3, provider comparison] The claim that 'Gmail being slightly more energy-efficient and Proton consuming more energy, particularly in memory usage' should be accompanied by the actual energy values in Joules and their spread, so that 'slightly' is quantified.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: the user-side measurement chain is empirical and independent of its conclusions.

full rationale

The paper's concrete empirical claim—that Gmail is slightly more energy-efficient and Proton consumes more energy, particularly in memory—is the output of Selenium-automated functional-unit runs monitored by the Green Metric Tool. Nothing in this chain defines the conclusion in terms of the input or fits a parameter to the data and then repackages it as a prediction. The tool's container-based energy attribution is asserted rather than validated against a power meter, but that is a measurement-validity concern, not a circularity: the energy readings are independent inputs, and the provider ranking is an empirical consequence rather than an input. The only prior numeric inputs are the per-GB network and server emission factors taken from Malmodin et al. (reference [10]), which are external literature values and are not fitted to, or derived from, the user-side measurements. References to the Green Coding Project's GMT documentation and Energy ID are tool pointers, not self-citations that carry a load-bearing theorem; the one author-owned artifact link is a code repository for reproducibility. The paper explicitly states that its measurements 'does not yet conclusively support or falsify our hypothesis regarding primary impacts,' so the central hypothesis is not forced. No equation or construction in the paper shows X defined in terms of Y, no fitted parameter is renamed as a prediction, and no uniqueness claim is imported from the authors' prior work. The identified weakness—unvalidated container-level energy attribution—is best treated as a correctness or external-validity risk, not as circular reasoning.

Assumptions & free parameters 0 free parameters · 4 assumptions · 0 invented entities

The central measurement rests on tool-based energy attribution and external emission factors, both taken on faith from prior work. No free parameters are fitted in this paper; the per-GB network/server factors and grid-mix conversion are external inputs.

assumptions (4)

domain assumption Green Metric Tool container isolation accurately attributes machine energy to the monitored web service.
Section 3 relies on the tool's claim that a container produces a controlled environment that 'accurately evaluate[s] the specific energy consumption and data transfer attributable to the container'; no validation against a physical power meter is provided.
domain assumption Network and server emissions scale with data volume according to sector-average per-GB factors.
Section 2.2 and 3.1 use Malmodin et al. ratios, while Section 3.1 acknowledges that network energy is not elastic, making linear allocation a simplifying assumption.
domain assumption Gmail and Outlook represent advertising- and tracking-financed services, and Proton Mail represents a privacy-preserving freemium service.
Section 1 assigns these business models and uses them to frame the hypothesis test; the classification is not independently verified.
domain assumption Standardized wait statements make cross-provider functional units comparable.
Section 3.1 says waits ensure consistency but reduce click rate and may hide time savings from ad-blocking, so comparability is bought at the cost of ecological validity.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Exploring Privacy and Security as Drivers for Environmental Sustainability in Cloud-Based Office Solutions (Extended Abstract)." pith.science (2026). https://pith.science/paper/D4YAFCL4

@misc{pith2026241116340,
  author       = {Pith},
  title        = {Pith review of: Exploring Privacy and Security as Drivers for Environmental Sustainability in Cloud-Based Office Solutions (Extended Abstract)},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/D4YAFCL4}},
  note         = {Machine review of arXiv:2411.16340}
}

read the original abstract

This paper explores the intersection of privacy, cybersecurity, and environmental impacts, specifically energy consumption and carbon emissions, in cloud-based office solutions. We hypothesise that solutions that emphasise privacy and security are typically "greener" than solutions that are financed through data collection and advertising. To test our hypothesis, we first investigate how the underlying architectures and business models of these services, e.g., monetisation through (personalised) advertising, contribute to the services' environmental impact. We then explore commonly used methodologies and identify tools that facilitate environmental assessments of software systems. By combining these tools, we develop an approach to systematically assess the environmental footprint of the user-side of online services, which we apply to investigate and compare the influence of service design and ad-blocking technology on the emissions of common web-mail services. Our measurements of a limited selection of such services does not yet conclusively support or falsify our hypothesis regarding primary impacts. However, we are already able to identify the greener web-mail services on the user-side and continue the investigation towards conclusive assessment strategies for online office solutions.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

18 extracted references · 16 canonical work pages

[1]

European Commission’s use of Microsoft 365 infrin ges data protec- tion law for EU institutions and bodies | European Data Prote ction Supervisor

2024. European Commission’s use of Microsoft 365 infrin ges data protec- tion law for EU institutions and bodies | European Data Prote ction Supervisor. https://www.edps.europa.eu/press-publications/press-news/press-releases/2024/european-commissions-

work page 2024
[2]

Software Freedom Conservancy. 2024. Selenium. https:/ /www.selenium.dev Accessed: 2024-10-02

work page 2024
[3]

Blair, and Adrian Friday

Charlotte Freitag, Mike Berners-Lee, Kelly Widdicks, B ran Knowles, Gordon S. Blair, and Adrian Friday. 2021. The real climate and transfo rmative impact of ICT: A critique of estimates, trends, and regulations. Patterns 2, 9 (Sept. 2021), 100340. https://doi.org/10.1016/j.patter.2021.100340

arXiv 2021
[4]

Vivian Frick, Ellen Matthies, John Thøgersen, and Tilma n Santarius. 2021. Do online environments promote suﬃciency or overconsumption ? Online adver- tisement and social media eﬀects on clothing, digital devic es, and air travel 3 LOCO ’24, December 03, 2024 Kayembe et al. consumption. Journal of Consumer Behaviour 20, 2 (March 2021), 288–308. https://do...

work page doi:10.1002/cb.1855 2021
[5]

JuiHsi Fu, PoChing Lin, and SingLing Lee. 2014. Detectin g spamming activi- ties in a campus network using incremental learning. Journal of network and computer applications 43 (2014), 56–65

work page 2014
[6]

Google. 2024. Gmail. https://www.google.com/gmail/a bout/ Accessed: 2024- 10-02

work page 2024
[7]

Green Coding Project. 2024. Philosophy and Methodology of Green Cod- ing. https://docs.green-coding.io/docs/prologue/phil osophy-methodology/ Ac- cessed: 2024-10-02

work page 2024
[8]

Green Coding Project. 2024. Philosophy and Methodology of Green Coding. https://www.green-coding.io/projects/energy-id/ Acce ssed: 2024-10-02

work page 2024

Show all 18 references

[9]

Jason Kayembe. 2024. Artifact: Exploring Privacy and Se curity as Dri- vers for Environmental Sustainability in Cloud-Based Oﬃce Solutions. https://github.com/jkayembe/thesis

2024
[10]

Jens Malmodin, Nina Lövehagen, Pernilla Bergmark, and Dag Lundén. 2024. ICT sector electricity consumption and greenhouse gas emissio ns–2020 outcome. Telecommunications Policy 48, 3 (2024), 102701

2024
[11]

Microsoft. 2024. Microsoft Outlook. https://outlook .live.com/mail/ Accessed: 2024-10-02

2024
[12]

Joshua M Pearce. 2020. Energy conservation with open so urce ad blockers. Tech- nologies 8, 2 (2020), 18

2020
[13]

Fabio Pesari, Giovanni Lagioia, and Annarita Paiano. 2 023. Client-side energy and GHGs assessment of advertising and tracking in the news w ebsites. Journal of Industrial Ecology 27, 2 (2023), 548–561

2023
[14]

PiHole. 2024. Overview of Pi-hole - Pi-hole documentat ion. https://docs.pi-hole.net/

2024
[15]

Proton. 2024. Proton Mail. https://proton.me/mail Ac cessed: 2024-10-02

2024
[16]

Pärssinen, M

M. Pärssinen, M. Kotila, R. Cuevas, A. Phansalkar, and J . Manner. 2018. Environ- mental impact assessment of online advertising. Environmental Impact Assess- ment Review 73 (Nov. 2018), 177–200. https://doi.org/10.1016/j.eiar .2018.08.004

2018 doi
[17]

Statista. 2024. Daily Number of Emails Worldwide from 2 017 to 2025. https://www.statista.com/statistics/456500/daily-number-of-e-mails-worldwide/ Accessed: 2024-10-02

2024
[18]

Statista. 2024. Global spam volume as percent- age of total e-mail traﬃc from 2011 to 2023. https://www.statista.com/statistics/420400/spam-ema il-traﬃc-share-annual/ Accessed: 2024-10-02. 4

2024

Pith tools

Reviewed August 12, 2026 · model on record in the stance chip above.

[1] [1]

European Commission’s use of Microsoft 365 infrin ges data protec- tion law for EU institutions and bodies | European Data Prote ction Supervisor

2024. European Commission’s use of Microsoft 365 infrin ges data protec- tion law for EU institutions and bodies | European Data Prote ction Supervisor. https://www.edps.europa.eu/press-publications/press-news/press-releases/2024/european-commissions-

work page 2024

[2] [2]

Software Freedom Conservancy. 2024. Selenium. https:/ /www.selenium.dev Accessed: 2024-10-02

work page 2024

[3] [3]

Blair, and Adrian Friday

Charlotte Freitag, Mike Berners-Lee, Kelly Widdicks, B ran Knowles, Gordon S. Blair, and Adrian Friday. 2021. The real climate and transfo rmative impact of ICT: A critique of estimates, trends, and regulations. Patterns 2, 9 (Sept. 2021), 100340. https://doi.org/10.1016/j.patter.2021.100340

arXiv 2021

[4] [4]

Vivian Frick, Ellen Matthies, John Thøgersen, and Tilma n Santarius. 2021. Do online environments promote suﬃciency or overconsumption ? Online adver- tisement and social media eﬀects on clothing, digital devic es, and air travel 3 LOCO ’24, December 03, 2024 Kayembe et al. consumption. Journal of Consumer Behaviour 20, 2 (March 2021), 288–308. https://do...

work page doi:10.1002/cb.1855 2021

[5] [5]

JuiHsi Fu, PoChing Lin, and SingLing Lee. 2014. Detectin g spamming activi- ties in a campus network using incremental learning. Journal of network and computer applications 43 (2014), 56–65

work page 2014

[6] [6]

Google. 2024. Gmail. https://www.google.com/gmail/a bout/ Accessed: 2024- 10-02

work page 2024

[7] [7]

Green Coding Project. 2024. Philosophy and Methodology of Green Cod- ing. https://docs.green-coding.io/docs/prologue/phil osophy-methodology/ Ac- cessed: 2024-10-02

work page 2024

[8] [8]

Green Coding Project. 2024. Philosophy and Methodology of Green Coding. https://www.green-coding.io/projects/energy-id/ Acce ssed: 2024-10-02

work page 2024