--- title: "CRYSTAL BALL: Checking Predictions Against the Record" document: P4047R0 date: 2026-05-01 audience: WG21 reply-to: - "Vinnie Falco " paper-type: informational --- ## Abstract This paper collects dated, public, falsifiable predictions about `std::execution` from proponents and critics alike, and checks each against the record. ## Revision History ### R0: May 2026 (pre-Brno mailing) * Initial revision. ### 1. Disclosure The author provides information and serves at the pleasure of the committee. This paper examines the published record. That effort requires re-examining consequential papers, including papers written by people the author respects. The author has papers before the committee proposing coroutine-based I/O ([P4003R3](https://isocpp.org/files/papers/P4003R3.pdf)) and analyzing `std::execution` ([P2583R4](https://isocpp.org/files/papers/P2583R4.pdf), [P4007R3](https://isocpp.org/files/papers/P4007R3.pdf), [P4014R2](https://isocpp.org/files/papers/P4014R2.pdf)). Every prediction and outcome cited below is sourced to public committee records, published blog posts, or recorded conference talks. **The author is the intelligence of record.** [P4023R0](https://www.open-std.org/jtc1/sc22/wg21/docs/papers/2026/p4023r0.pdf) [1] (Directions Group, "Strategic Direction for AI in C++") establishes that "the ultimate responsibility for accuracy, logic, and normative quality rests entirely with the human author." This paper follows that principle. AI tools assist with research, compilation, and drafting. The author curates, verifies, and takes responsibility for every claim. **This paper uses AI at every stage.** Prediction discovery, outcome verification, scorecard organization, and paper drafting were all AI-assisted. Every stage involves machine output. P4023R0 identifies research, summarization, and consistency checking as permitted uses of AI in the committee process [1]. This paper's use of AI falls within that scope. **Human curation is required at every stage.** AI can misidentify predictions, misclassify outcomes, or miss relevant public statements. The author verified every prediction-outcome pair against the cited source. **Outcome classifications reflect the author's assessment of the cited evidence.** The scorecard is one systematic reading of the public record. Each entry cites its source. Readers who disagree with a classification can verify against the cited source and reach their own conclusion. This paper asks for nothing. ### 2. Method A prediction qualifies if it is (a) dated, (b) public, and (c) falsifiable - a claim about a future outcome that can be checked. Outcomes are graded: **Confirmed,** **Unconfirmed,** **Partially** Confirmed, or **Overtaken by** Events. Both sides miss. Both sides hit. The table includes all of them. ### 3. The Predictions #### Timeline **Source** **Date** **Outcome** T1 P2300 unlikely for C++23 given size of the proposal Adelstein Lelbach, CppCon 2021 [2] 2021 Confirmed T2 "merge most of the foundational bits [P2300] early in the C++26 cycle" Voutilainen 2022 Confirmed T3 "not going to be in a timeframe conducive to inclusion in C++23" (re: language changes P2300 needs) [P2469R0](https://www.open-std.org/jtc1/sc22/wg21/docs/papers/2021/p2469r0.pdf) [4] 2021-10 Confirmed T4 Networking TS should ship for C++23 [P2469R0](https://www.open-std.org/jtc1/sc22/wg21/docs/papers/2021/p2469r0.pdf) [4] 2021-10 Unconfirmed - neither Net TS nor any networking shipped T5 Stop Networking TS for C++23, focus on P2300 Voutilainen 2021 Confirmed - Net TS stopped; P2300 proceeded #### Safety and Correctness **Source** **Date** **Outcome** S1 Stack overflow from reentrant completion without executors is "user-hostile default behaviour" [4] 2021-10 Confirmed - Müller [P3801R0](https://www.open-std.org/jtc1/sc22/wg21/docs/papers/2025/p3801r0.html) [6] (2025), Kühl [P3796R0](https://www.open-std.org/jtc1/sc22/wg21/docs/papers/2025/p3796r1.html) [7] S2 P2300 design "insufficiently paranoid to be suited to internet-facing software development" [8] 2021-10 Partially confirmed - networking deferred to C++29 S3 `set_error` channel confusion - scheduling errors vs. I/O errors [4] 2021-10 Confirmed - still unresolved S4 "P2300 will likely require language changes to work effectively" - tail calls, trampolines [4] 2021-10 Confirmed - [P2583R4](https://isocpp.org/files/papers/P2583R4.pdf) [9] documents symmetric transfer gap (2026) S5 Tail call concerns documented before P2300 Nishanov [10] 2019 Confirmed - still open #### Customization Mechanism **#** **Prediction** **Source** **Date** **Outcome** C1 `tag_invoke` as the customization mechanism for sender algorithms [11] 2020-2022 Unconfirmed - abandoned; replaced by `transform_sender` ([P2999R3](https://www.open-std.org/jtc1/sc22/wg21/docs/papers/2023/p2999r3.html) [12]) C2 `transform_sender` as the customization mechanism [12] 2024 Unconfirmed - [P3303R1](https://www.open-std.org/jtc1/sc22/wg21/docs/papers/2024/p3303r1.html) [13] "Fixing," [P3718R0](https://www.open-std.org/jtc1/sc22/wg21/docs/papers/2025/p3718r0.html) [14] "Fixing Again," then [P3826R5](https://www.open-std.org/jtc1/sc22/wg21/docs/papers/2026/p3826r5.html) [15] proposes removal C3 "the customization mechanism has seen a fair bit of recent churn" - remove for C++26, restore in C++29 Niebler [15] 2025-10 Pending - proposed #### Universality and Scope **Source** **Date** **Outcome** U1 ">90% of all async code in the future should be coroutines" Niebler, "Structured Concurrency" [16] 2020-11 Shifted - by 2024: "returning a sender is a great choice" [17] U2 "senders are like iterators when STL emerged" - universal adoption predicted Kühl, quoted in 2021-25 Unconfirmed - no equivalent adoption curve U3 SG1 "no consensus" that one grand unified model is needed (4-9-5-5-1) [P2453R0](https://www.open-std.org/jtc1/sc22/wg21/docs/papers/2022/p2453r0.html) [18] 2021-09 Confirmed - poll happened; direction proceeded regardless U4 "Networking should be based on sender/receiver model" - weak consensus (17-11-10-4-6) [P2453R0](https://www.open-std.org/jtc1/sc22/wg21/docs/papers/2022/p2453r0.html) [18] 2021-10 Unconfirmed - no sender-based networking exists; deferred to C++29 U5 Production library chose awaitables over senders for usability Prokoptsev, CppCon 2024 [19] 2024 Confirmed U6 "C++26 will *still* not have a standard coroutine task type" Niebler 2025-10 Pending #### Domain Viability **#** **Prediction** **Source** **Date** **Outcome** D1 Sender/receiver deployed in shipping Facebook products, "monthly users number in the billions" Niebler [P2470R0](https://www.open-std.org/jtc1/sc22/wg21/docs/papers/2021/p2470r0.pdf) [20] 2021 Confirmed D2 NVIDIA "fully invested in P2300... we plan to ship in production sometime next year" NVIDIA statement in [P2470R0](https://www.open-std.org/jtc1/sc22/wg21/docs/papers/2021/p2470r0.pdf) [20] 2021 Confirmed - nvexec shipped D3 Bloomberg experimentation with sender/receiver [P2470R0](https://www.open-std.org/jtc1/sc22/wg21/docs/papers/2021/p2470r0.pdf) [20] 2021 Confirmed #### Networking **Source** **Date** **Outcome** N1 "The C++ Standard can't access the Internet. This is embarrassing and long overdue." [4] 2021-10 Confirmed - still true N2 "We don't have a composable approach proposal [for networking] at this point" Voutilainen [3] 2022 Confirmed - still true N3 Chair's note: "In the short term, this poll result doesn't mean much. We don't have a paper in hand that proposes networking based on the P2300 model." [18] 2021-10 Confirmed - still no such paper #### Implementation Maturity **#** **Prediction** **Source** **Date** **Outcome** M1 "many algorithms clearly labeled 'Not yet implemented' in libunifex" [P2469R0](https://www.open-std.org/jtc1/sc22/wg21/docs/papers/2021/p2469r0.pdf) [4] 2021-10 Confirmed - libunifex archived; 50+ post-approval items in C++26 WD [21] M2 P2300 task type concerns: stack overflow, allocator, error handling Müller [P3801R0](https://www.open-std.org/jtc1/sc22/wg21/docs/papers/2025/p3801r0.html) [6], Kühl [P3796R1](https://www.open-std.org/jtc1/sc22/wg21/docs/papers/2025/p3796r1.html) [7] 2025 Confirmed - 16 open issues documented by task author ### 4. Scorecard **Category** **Predictions** **Confirmed** **Unconfirmed** **Shifted /** **Partial** **Pending** | Timeline | 5 | 4 | 1 | 0 | 0 | | --- | --- | --- | --- | --- | --- | | Safety / Correctness | 5 | 4 | 0 | 1 | 0 | | Customization Mechanism | 3 | 0 | 2 | 0 | 1 | | Universality / Scope | 6 | 2 | 2 | 1 | 1 | | Domain Viability | 3 | 3 | 0 | 0 | 0 | | Networking | 3 | 3 | 0 | 0 | 0 | | Implementation Maturity | 2 | 2 | 0 | 0 | 0 | | **Total** | **27** | **18** | **5** | **2** | **2** | ### 5. Conclusion | Predictive | Not Predictive | | --- | --- | | Safety and correctness concerns | Universality claims | | Domain viability (where deployed) | Customization mechanism design | | Networking gap | Timeline estimates (all sides) | Implementation maturity concerns ## References [1] Michael Garland, Andrew McKenney, Roger Orr, Bjarne Stroustrup, Daveed Vandevoorde, Michael Wong. [P4023R0](https://www.open-std.org/jtc1/sc22/wg21/docs/papers/2026/p4023r0.pdf). "Strategic Direction for AI in C++: Governance, and Ecosystem" (Directions Group, 2026). [2] [Bryce Adelstein Lelbach. "C++ Standard Parallelism." CppCon 2021](https://www.youtube.com/watch?v=LW_T2RGXego) [3] [Ville Voutilainen. P0592R5. "To boldly suggest an overall plan for C++26." WG21](https://www.open-std.org/jtc1/sc22/wg21/docs/papers/2022/p0592r5.html) [4] [Christopher Kohlhoff, Jamie Allsop, Vinnie Falco, Richard Hodges, Klemens Morgenstern. P2469R0. "Response to P2464:](https://www.open-std.org/jtc1/sc22/wg21/docs/papers/2021/p2469r0.pdf) [The Networking TS is baked, P2300 Sender/Receiver is not." WG21, 2021](https://www.open-std.org/jtc1/sc22/wg21/docs/papers/2021/p2469r0.pdf) [5] [Ville Voutilainen. P2464R0. "Ruminations on networking and executors." WG21, 2021](https://www.open-std.org/jtc1/sc22/wg21/docs/papers/2021/p2464r0.html) [6] [Jonathan Müller. P3801R0. "Concerns about the design of std::execution::task." WG21, 2025](https://www.open-std.org/jtc1/sc22/wg21/docs/papers/2025/p3801r0.html) [7] [Dietmar Kühl. P3796R1. "Coroutine Task Issues." WG21, 2025](https://www.open-std.org/jtc1/sc22/wg21/docs/papers/2025/p3796r1.html) [8] [Christopher Kohlhoff, Jamie Allsop, Klemens Morgenstern. P2480R0. "Response to P2471: "NetTS, Asio, and Sender library](https://www.open-std.org/jtc1/sc22/wg21/docs/papers/2021/p2480r0.pdf) [design comparison" - corrected and expanded" WG21, 2021](https://www.open-std.org/jtc1/sc22/wg21/docs/papers/2021/p2480r0.pdf) [9] [Mungo Gill, Vinnie Falco. P2583R4. WG21, 2026](https://isocpp.org/files/papers/P2583R4.pdf) [10] [Gor Nishanov. P1362R0. "Incremental Approach: Coroutine TS + Core Coroutines." WG21, 2019](https://www.open-std.org/jtc1/sc22/wg21/docs/papers/2018/p1362r0.pdf) [11] [Michael Garland, et al. P2403R0. "Presentation on P2300 - std::execution." WG21](https://www.open-std.org/jtc1/sc22/wg21/docs/papers/2021/p2403r0.pdf) [12] [Eric Niebler. P2999R3. "Sender Algorithm Customization." WG21, 2024](https://www.open-std.org/jtc1/sc22/wg21/docs/papers/2023/p2999r3.html) [13] [Eric Niebler. P3303R1. "Fixing Lazy Sender Algorithm Customization." WG21, 2024](https://www.open-std.org/jtc1/sc22/wg21/docs/papers/2024/p3303r1.html) [14] [Eric Niebler. P3718R0. "Fixing Lazy Sender Algorithm Customization, Again." WG21, 2025](https://www.open-std.org/jtc1/sc22/wg21/docs/papers/2025/p3718r0.html) [15] [Eric Niebler. P3826R5. "Fix Sender Algorithm Customization" WG21, 2025](https://www.open-std.org/jtc1/sc22/wg21/docs/papers/2026/p3826r5.html) [16] [Eric Niebler. "Structured Concurrency." 2020](https://ericniebler.com/2020/11/08/structured-concurrency/) [17] [Eric Niebler. "What are Senders Good For, Anyway?" 2024](https://ericniebler.com/2024/02/04/what-are-senders-good-for-anyway/) [18] [Bryce Adelstein Lelbach. P2453R0. "2021 October Library Evolution Poll Outcomes." WG21, 2022](https://www.open-std.org/jtc1/sc22/wg21/docs/papers/2022/p2453r0.html) [19] [Dmitry Prokoptsev. "C++ Coroutines and Structured Concurrency in Practice." CppCon 2024](https://www.youtube.com/watch?v=aPqMQ7SXSiw) [20] [Eric Niebler. P2470R0. "Slides for presentation of P2300R2: std::execution (sender/receiver)" WG21, 2021](https://www.open-std.org/jtc1/sc22/wg21/docs/papers/2021/p2470r0.pdf) [21] [LWG unresolved prioritized list](https://cplusplus.github.io/LWG/unresolved-prioritized.html) [22] [P2300R10](https://www.open-std.org/jtc1/sc22/wg21/docs/papers/2024/p2300r10.html) - "`std::execution`" (Eric Niebler, Michał Dominiak, Georgy Evtushenko, Lewis Baker, Lucian Radu Teodorescu, Lee Howes, Kirk Shoop, Michael Garland, Bryce Adelstein Lelbach, 2024).