Skip to content

Inaccuracies in newsletter "State of Rust survey findings" #2

@apiraino

Description

@apiraino

Hi,

I am a contributor of the Rust project and in our newsletter (This Week in Rust) I've found a link to your latest newsletter:
https://weeklyrust.substack.com/p/state-of-rust-survey-findings

I am also co-maintainer of the Rust Survey and have authored the blog post you mention in your findings article. I think there are a few inaccuracies or biased representions:

Production use keeps climbing. 38% of respondents report significant Rust usage at their organizations, and 25 % expect their company to hire Rust developers next year. Web services (60 %), embedded (50 %), and CLI tools (48 %) lead.

Unsure where these numbers come from: if they refer to the question "Are you personally using Rust at work?", the correct number should be that almost 40% use Rust regularly at work ("Yes, for the majority of my coding").

"Web Services" (which I assume you mean by coalescing server-side and cloud computing applications) amount for about 53% and 52%, see graph for "In what technology domain(s) is Rust used at your organisation?", embedded use exceed 50% when adding all embedded areas in the graph.

Learning resources are excellent: 98 % rely on the official docs, 58 % on books, and 49 % on crate source code

86% of the respondents rely on official documentation, 84.7% on books etc. ... those numbers look incorrect

Rust’s learning curve remains a barrier. Among the 358 respondents who don’t use Rust, 22 % called Rust “too difficult to learn,” with word-clouds full of “syntax,” “confusing,” and even “toxic.”

Clarification: the word "Toxic" in the wordcloud for "Why don't you use Rust?" is very small and comes from that 13.2% of "other" answers. Comparing to other objective hurdles like "being complex" (31.7%) doesn't really give justice to the proportions. I didn't find (or maybe I just missed) the word "confusing" you quoted.

Compile times and disk usage remain major complaints: 17% say slow compilation is a serious problem, and 34% say it could be significantly better.

Yes, compile times are the #1 complaint but the percentage we publish is 27.2% and 54.7% declare that "could be improved, but does not limit me"

Formal education remains rare. 94.5% haven’t taken a Rust course in the last year.

Unsure why this is a concern. Also considering that +61% of our respondents state that they have +10 years of programming experience :-) Besides "94.5%" is from 2024 , for 2025 the percentage was 94.8%. Small difference, just missed the right column.

Enterprise adoption still feels hesitant: 36 % of organizations have no plans to hire Rust devs, and 29 % simply don’t know.

The graph for "Is your organisation planning on hiring Rust developers in the next year?" shows that 19.5% of organizations not planning to hire Rust experts

Deeper community involvement is low, 77 % have never attended a Rust conference ...

Unsure where this number comes from

... and 87 % have never contributed to an RFC.

This is a piece of data that without context shows a wrong bias. Contributing an RFC is a quite specialized work that often requires domain knowledge. Not contributing an RFC carries no meaning per se.

[...] fix the learning curve and compile times, and Rust could go from popular to dominant.

Well, that's a bold statement, we in the Rust project really wish it was that simple :-)


May I ask if you used some LLM to write that article? Asking because since they are not able to reason about data, they sometimes mixup facts and numbers.

As I read that you're happy to listen to feedback, I hope this feedback will be somewhat helpful.

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions