pith. sign in

arxiv: 1906.10398 · v1 · pith:MYEKYZPXnew · submitted 2019-06-25 · 💻 cs.CY

Future of Computing is Boring (and that is exciting!) or How to get to Computing Nirvana in 20 years or less

Aleksander Slominski , Vinod Muthusamy , Vatche Ishakian This is my paper

Pith reviewed 2026-05-25 16:30 UTC · model grok-4.3

classification 💻 cs.CY
keywords cloud computingutility computingdeveloper productivitycost trendsstandardizationmetered servicesfuture of computing
0
0 comments X

The pith

Saving developer time is the key priority for future computing because human effort now dominates over falling compute costs.

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

The paper traces computing's path toward becoming a metered utility like the electric grid, where economies of scale and standardization replace local setups. It contrasts sharply falling compute costs with the steady high cost of human developers to conclude that designs must focus on minimizing developer effort. A sympathetic reader would care because this cost shift implies computing will be delivered as a background service rather than something teams build and manage directly. The authors outline remaining steps to complete this evolution and reach a seamless state in 20 years or less.

Core claim

Computing follows the electric grid's trajectory from local generation to centralized, standardized utility provision, but this process remains unfinished; because the cost of human developer time now greatly exceeds the cost of compute resources, the dominant task ahead is to create systems and abstractions that save developer time.

What carries the argument

The cost-ratio comparison between compute resources and human developer time that identifies human time as the dominant ongoing constraint.

If this is right

  • Economies of scale will continue moving computing from local and scientific grids to commercial metered cloud utilities.
  • Standardization will reduce custom infrastructure work and favor managed services over bespoke setups.
  • Future systems will emphasize high-level interfaces that hide complexity so programmers spend less time on operations.
  • The end state is computing that users access as routinely as electricity without managing the underlying grid.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • If the cost ratio holds, research and investment will naturally flow toward abstraction layers and automation that multiply developer output.
  • Adoption of the utility model would require parallel solutions for security, privacy, and reliability at the provider level.
  • Educational programs may shift emphasis from building low-level systems to composing and extending utility services.

Load-bearing premise

The cost of human developer time will remain high relative to compute costs and no major technological or economic disruption will alter this ratio over the next twenty years.

What would settle it

Empirical data showing either a steep sustained drop in skilled developer compensation or a technology that makes equivalent output possible with far less human input within the next twenty years.

read the original abstract

We see a trend where computing becomes a metered utility similar to how the electric grid evolved. Initially electricity was generated locally but economies of scale (and standardization) made it more efficient and economical to have utility companies managing the electric grid. Similar developments can be seen in computing where scientific grids paved the way for commercial cloud computing offerings. However, in our opinion, that evolution is far from finished and in this paper we bring forward the remaining challenges and propose a vision for the future of computing. In particular we focus on changes in cost of computing and high cost of human time in comparison that indicates that saving developer time is the most important for future of computing.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

1 major / 1 minor

Summary. The paper argues that computing is evolving toward a metered utility model analogous to the electric grid, with scientific grids leading to commercial clouds; it identifies remaining challenges and proposes a vision for the next 20 years in which saving developer time is the dominant priority because human labor costs will continue to exceed and dominate falling compute costs.

Significance. If the cost-dominance premise were demonstrated, the vision could usefully redirect research agendas in cloud systems and software engineering toward developer-productivity tools and automation. The manuscript offers no quantitative trends, cost ratios, or falsifiable predictions, however, so its influence on the field would remain speculative.

major comments (1)
  1. [Abstract] Abstract: the claim that 'changes in cost of computing and high cost of human time in comparison that indicates that human time is the dominant constraint' is asserted without any cost figures, historical ratios, citations to economic studies, or sensitivity analysis. This unsupported premise is load-bearing for the conclusion that 'saving developer time is the most important for future of computing.'
minor comments (1)
  1. [Abstract] Abstract: the phrasing 'high cost of human time in comparison that indicates' is grammatically unclear and should be reworded for precision.

Simulated Author's Rebuttal

1 responses · 0 unresolved

We thank the referee for the constructive feedback. We address the major comment point-by-point below and agree that additional supporting material will strengthen the manuscript.

read point-by-point responses

  1. Referee: [Abstract] Abstract: the claim that 'changes in cost of computing and high cost of human time in comparison that indicates that human time is the dominant constraint' is asserted without any cost figures, historical ratios, citations to economic studies, or sensitivity analysis. This unsupported premise is load-bearing for the conclusion that 'saving developer time is the most important for future of computing.'

    Authors: We agree that the abstract presents this premise without quantitative backing or citations, and that the claim is central to the paper's argument. As the manuscript is a forward-looking vision piece, it prioritizes implications over new empirical analysis. However, the referee's point is valid: adding context will make the premise more robust. In the revised version we will (1) include brief historical cost trends and ratios drawn from existing literature (e.g., hardware price/performance declines and software labor cost studies), (2) add 2-3 key citations to economic analyses of IT costs, and (3) revise the abstract wording to reference these trends without changing the overall vision. This directly addresses the load-bearing concern while preserving the paper's speculative character. revision: yes

Circularity Check

0 steps flagged

No circularity: vision paper states premise without self-referential derivation

full rationale

The provided text contains no equations, fitted parameters, self-citations, or derivation steps that reduce a claimed result to its own inputs by construction. The central assertion—that changes in computing costs versus high human time costs indicate developer-time savings as the dominant future priority—is presented directly as an opinion based on observed trends, without any mathematical or definitional loop. No load-bearing uniqueness theorem, ansatz smuggling, or renaming of known results occurs. The paper is a forward-looking vision piece whose claims stand or fall on external economic data rather than internal circular construction.

Axiom & Free-Parameter Ledger

0 free parameters · 2 axioms · 0 invented entities

The argument depends on two untested domain assumptions about persistent cost ratios and the uninterrupted continuation of the utility model; no free parameters or invented entities are introduced because the text contains no quantitative model.

axioms (2)
  • domain assumption The relative cost of human developer time versus compute will remain high and stable for the next two decades.
    Invoked to conclude that developer time is the primary optimization target.
  • domain assumption The evolution of computing will follow the same utility-grid trajectory as electricity without major interruptions.
    Underpins the claim that the transition is 'far from finished' and will reach a defined end state.

pith-pipeline@v0.9.0 · 5653 in / 1203 out tokens · 36522 ms · 2026-05-25T16:30:17.120947+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

15 extracted references · 15 canonical work pages

  1. [1]

    A computational grid is a hardware and software infrastructure that provides dependable, consistent, pervasive, and inexpensive access to high-end computational capabilities

    The Future of Computing is Boring (and that is exciting!) How to get to Computing Nirvana in 20 years or less? Aleksander Slominski IBM T.J Watson Research Center New York, NY aslom@us.ibm.com Vinod Muthusamy IBM T.J Watson Research Center New York, NY vmuthus@us.ibm.com Vatche Ishakian ​ Computer Information Systems Bentley University, Waltham, MA vishak...

    work page 2006

  2. [2]

    Containers provide standardized packaging (into container images) and isolation for processes that are running in a containerized environment on top of one shared (virtual) machine instead of creating a VM for each computation (see the section below for more discussion about rise of containers). Kubernetes added orchestration for containers in 2014 to pro...

    work page 2014

  3. [3]

    ** Main advance in computation are by combining CPU with GPUs

    $0.02** $7 * This is an extrapolated cost as there were hardware limitation on getting any machine with that level of performance. ** Main advance in computation are by combining CPU with GPUs. Table II shows how much computation (GFLOPS) we can buy with $100K. We also show the ratio of change: we can see that the rate of change slowed down for memory and...

    work page 2017

  4. [4]

    However, that is not true about human resources as developer’s time is getting more expensive when compared to how much compute time can be bought

    5M GFLOPS or 5 PFLOPS (100x) 15000 GB or 15 TB (3x) We can see several orders of magnitude more computational resources can be bought with the same amount of money over a 10 to 20-year span. However, that is not true about human resources as developer’s time is getting more expensive when compared to how much compute time can be bought. In our comparison,...

    work page 1960

  5. [5]

    Hourglass for future computing with containers serving as one shared standard The container standardization should be rich enough to run the majority of today’s applications. To support legacy computing code, it may be required to support running old computing non-standardized containerization formats (such as VMs) wrapped as a container inside a standard...

    work page 2013

  6. [6]

    The Grid: Blueprint for a New Computing Infrastructure,

    [GridBook] C. Kesselman and I. Foster, “The Grid: Blueprint for a New Computing Infrastructure,” Morgan Kaufmann Publishers Inc., San Francisco, CA, USA, 1999

    work page 1999

  7. [7]

    Client statistics by OS

    [FoldingAtHomeStats] “Client statistics by OS”, visited 2018-11-23 https://stats.foldingathome.org/os

    work page 2018

  8. [8]

    A View of Cloud Computing,

    [CloudVision] M. Armbrust, A. Fox, R. Grith, A. D. Joseph, R. Katz, A. Konwinski, G. Lee, D. Patterson, A. Rabkin, I. Stoica and M. Zaharia, “A View of Cloud Computing,” Communications of the ACM, Vol. 53, No. 4, 2010, pp. 50-58. http://dx.doi.org/10.1145/1721654.1721672

    work page doi:10.1145/1721654.1721672 2010

  9. [9]

    Software Developer Statistics: How Many Software Engineers Are There in the US and the World?

    [DevelopersWorldwide] “Software Developer Statistics: How Many Software Engineers Are There in the US and the World?”, Oct 31, 2017, https://www.daxx.com/article/software-developer-statistics-2017-pro grammers

    work page 2017

  10. [10]

    Salary Trends Through Salary Survey: A Historical Perspective on Starting Salaries for New College Graduates

    [SalaryGraduates] “Salary Trends Through Salary Survey: A Historical Perspective on Starting Salaries for New College Graduates”, August 02, 2016 http://www.naceweb.org/job-market/compensation/salary-trends-thro ugh-salary-survey-a-historical-perspective-on-starting-salaries-for-ne w-college-graduates/

    work page 2016

  11. [11]

    Cost of computing - Approximate cost per GFLOP

    [GflopsCost] “Cost of computing - Approximate cost per GFLOP”, November 2018, https://en.wikipedia.org/wiki/FLOPS#Hardware_costs

    work page 2018

  12. [12]

    [MemoryPrices] ”Memory Prices (1957-2018)” https://jcmit.net/memoryprice.htm

    work page 1957

  13. [13]

    Comparing Cloud Instance Pricing: AWS vs Azure vs Google vs IBM

    [RightScaleCompare] “Comparing Cloud Instance Pricing: AWS vs Azure vs Google vs IBM”, November 18, 2017, https://www.rightscale.com/blog/cloud-cost-analysis/comparing-clou d-instance-pricing-aws-vs-azure-vs-google-vs-ibm

    work page 2017

  14. [14]

    The future of Linux Containers

    [Docker] “The future of Linux Containers”, 2013 https://www.youtube.com/watch?v=wW9CAH9nSLs

    work page 2013

  15. [15]

    After Moore's Law

    [MooresLaw] T. Cross "After Moore's Law" The Economist Technology, 2016-03 https://www.economist.com/technology-quarterly/2016-03-12/after-m oores-law

    work page 2016