TL;DR
Get the latest gadgets delivered free with Prime
- Fast, free delivery on millions of items
- Prime Video, Amazon Music and more included
- Member-only deals all year
The Pragmatic Engineer published an interview with Sean Allen, a distinguished engineer at Antithesis, on software correctness, testing, programming languages and supply chain security. Allen argues that correctness should be designed into systems from the start and says there is no single, simple defense against malicious package attacks.
The Pragmatic Engineer has published an interview with Sean Allen, a distinguished engineer at software testing company Antithesis, on how developers can build more correct systems and why software supply chain attacks have no simple fix. Allen argues that teams should prevent invalid states through design and use repeated workloads to check system behavior, rather than relying on late-stage quality gates or manual testing alone.
Allen defines software correctness in practical terms: a program should do what its users and operators expect. He said engineers often lack a complete specification at the outset, even when they can recognize a failure. He frames those expectations as invariants—guarantees about how software behaves—and says software is correct enough when it serves its purpose without causing an outage that wakes someone in the night.
On testing, Allen challenges the idea that more manual checks are the best route to reliability. He favors designing systems so that illegal states cannot be represented in the first place, including through language and type-system choices. He also discusses running defined workloads repeatedly to examine behavior under stress. The source gives the Bank Test, used to stress database isolation guarantees, as an example; it does not provide performance results or a comparison with other testing methods.
Allen also warns that package attacks are difficult to prevent because malicious code can enter through a hijacked dependency and run on a machine or service. He says languages with stronger defensive properties, such as E, have not become popular, while widely used languages often assume a trustworthy software ecosystem. The interview presents this as his assessment, not as a finding from a new security study. Allen’s discussion ranges across the Pony programming language, garbage collection, technical debt and upfront design.
Designing Reliability Before Release
The interview speaks to engineering teams facing a practical tension: they need to ship software while limiting defects and security exposure. Allen’s approach puts correctness into design and development, rather than treating it as a final approval step. If teams can specify important invariants and exercise realistic workloads throughout development, they may find failures earlier; the interview does not quantify how much this reduces defects.
His supply chain warning also points to a risk beyond an individual company’s code. Applications depend on packages and tools produced elsewhere, and a compromised dependency can carry malicious behavior into a service. Allen’s position that no single defense is sufficient cautions against treating language choice, testing, or a release gate as a complete answer. The article offers engineering perspectives, not a prescribed security program.
Products Worth Considering
DJI Osmo Mobile 7P Gimbal Stabilizer with Mic Mini Transmitter, Infinity Black
Logitech MX Ergo S Advanced Wireless Trackball Mouse USB-C – Graphite
Logitech MX Master 4 Ergonomic Wireless Mouse with Haptic Feedback, Graphite
As an affiliate, we earn on qualifying purchases.
The Interview’s Engineering Focus
The interview was published by The Pragmatic Engineer and is presented as a conversation about modern software engineering. Its editor says the discussion was prompted by Allen’s work on confidential computing at Microsoft, his contributions to Pony, and his current focus on correctness at Antithesis. The source notes that Antithesis sponsors the publication’s podcast but says this interview was not sponsored by the company.
The publication connects the topic to the growing volume of code produced by AI agents, suggesting that more generated code could create more opportunities for bugs. That is the editor’s motivation for the conversation, not a measured claim about the amount of code being shipped or the number of resulting incidents. The interview also includes a separate announcement that The Pragmatic Summit is scheduled for Feb. 10 in San Francisco, with tickets listed at $1,000. That event notice is separate from Allen’s technical discussion.
““It’s down to whether my program does what I expect.””
— Sean Allen, distinguished engineer at Antithesis
Limits of the Interview’s Evidence
The source is an interview and does not include independent testing data comparing Allen’s preferred methods with manual testing, unit tests, or late-stage quality gates. It does not specify a particular package compromise, quantify supply chain incidents, or evaluate the effectiveness of named security controls. The discussion therefore records Allen’s engineering views rather than establishing that one approach works best across all systems.
It is also unclear from the supplied material which specific language features or implementation practices teams should adopt to reduce dependency risk. Allen’s comments about language popularity and ecosystem trust are broad observations in the conversation, not a complete account of available defenses. The source excerpt ends partway through a later response, so it does not contain the full interview.
Applying the Ideas in Practice
The source does not announce a product release, policy change, or follow-up study. For engineering teams, the next step implied by Allen’s advice is to identify the behaviors a system must preserve, express those as testable invariants where possible, and run relevant workloads repeatedly during development. That is an interpretation of the interview, not a formal recommendation or checklist published by Antithesis.
The Pragmatic Engineer says readers can watch the conversation on YouTube. The publication also says applications are open for its Feb. 10 San Francisco summit, but it provides no year in the supplied excerpt; the date and ticket details should be checked against the event listing before relying on them.
Key Questions
Who is Sean Allen?
Sean Allen is a distinguished engineer at Antithesis, a core contributor to the Pony programming language, and an engineer with experience in distributed systems, stream processing and confidential computing, according to the interview.
What does Allen mean by software correctness?
He describes correctness as a program doing what people expect it to do. He uses invariants to describe guarantees about that behavior, while noting that teams do not always know their full expectations before they encounter problems.
Why does Allen criticize manual testing?
Allen argues that preventing invalid states through software design and language features can catch some problems before testing. He also discusses repeatedly running workloads, but the interview does not claim that testing should be eliminated or provide a comparative evaluation of methods.
What is Allen’s view of supply chain security?
He says a hijacked package can introduce malicious code into a machine or service, and that there is no single, simple fix. His discussion points to challenges involving language defenses and assumptions of trust in software ecosystems; it does not set out a complete security plan.
Source: newsletter.pragmaticengineer.com
