The 500-Package App: Open Source Myths and the Number That Matters
Archi4 min read
Contents
- The 500-Package App and the Myths That Follow It
- Myth: Open source is free, so maintenance is free
- Myth: A license change is the biggest threat
- Myth: More stars means a healthier project
- Myth: An SBOM solves dependency risk
- Myth: Corporate backing covers the long tail
- Myth: Pinning versions removes the risk
The 500-Package App and the Myths That Follow It#
A mid-sized web application can carry roughly 500 open source packages in its lockfile. The exact number varies by stack, but the order of magnitude is what matters. Those 500 names arrive through direct dependencies and transitive ones, each with a license, a release history, and usually a small group of people who approve changes. The myths start when teams treat that list as a product catalog instead of a maintenance queue.
Myth: Open source is free, so maintenance is free#
Fact: The code has no license fee. The work does. Someone reviews pull requests, triages bug reports, cuts releases, and answers security questions. For many projects, that someone is a volunteer with a day job. For others, it is one paid engineer funded by grants or a company that uses the package. The 500-package app spreads that labor across hundreds of small arrangements. A team that budgets for cloud spend and none for upstream maintenance has misread its own supply chain.
The number to watch is not the 500 packages. It is how many of them have one active maintainer. A single person can keep a small parser alive for years. That works until a CVE arrives during a holiday, or the maintainer changes jobs. Companies do not need to fund all 500. They need to know which 20 or 30 carry the most risk.
Myth: A license change is the biggest threat#
Fact: License changes make headlines because they are visible. Operational neglect is quieter and more common. A package can keep the same permissive license while its release cadence stops, its CI breaks, and its issue tracker fills with unanswered reports. The lockfile still resolves. The build still passes. The risk sits in the gap between works today and will receive a patch next quarter.
License risk deserves review, especially for products that ship binaries or embed code in appliances. But a license audit does not tell you whether the maintainer has time to review a security fix. Treat legal review and maintenance review as separate checks.
Myth: More stars means a healthier project#
Fact: Stars measure attention, not maintenance. A project with 40,000 stars can have a two-person review queue and a backlog of unreleased fixes. A project with 300 stars can have a paid maintainer, a steady release train, and a security policy. The useful signals are less glamorous: time to merge a small fix, frequency of releases, whether security advisories get a response, and how many people can approve a change.
For the 500-package app, stars are also hard to collect at scale. A better exercise is to sort dependencies by release activity and count the ones with no release in two years. That list is shorter and more actionable than a popularity rank.
Myth: An SBOM solves dependency risk#
Fact: A software bill of materials is an inventory. It tells you what you have. It does not tell you who is maintaining it, whether the version is patched, or whether the build actually uses the vulnerable code path. Teams that generate an SBOM and file it away have created a receipt, not a process.
The useful workflow connects the inventory to decisions. When a package appears in the SBOM, the team should know its owner, its update path, and the test that proves an upgrade is safe. That sounds like ordinary engineering. It is. The SBOM is only the list of parts.
Myth: Corporate backing covers the long tail#
Fact: Corporate backing often concentrates on the projects a company ships or depends on directly. The long tail is everything else: the small date parser, the color library, the build plugin that saves a week of work. Those packages rarely have a foundation, a sponsor, or a security team. They persist because they do a job and do not ask for attention.
This is where the 500 number becomes useful. Divide the lockfile into three groups: packages the company would pay to keep alive, packages it could replace with a few days of work, and packages nobody has examined. The middle group is often larger than expected. Replacing a tiny dependency can be cheaper than funding a maintenance contract for it.
Myth: Pinning versions removes the risk#
Fact: Pinning gives reproducibility. It does not give immunity. A pinned vulnerable version stays vulnerable until someone updates it, tests it, and ships it. The lockfile can make a team feel controlled while a known issue sits in production for months. The useful practice is a schedule: review advisories, test upgrades, and keep the path from upstream release to deployed build short. If the path takes a quarter, pinning has hidden the risk rather than removed it. The 500 names in the lockfile are not a set of museum pieces. They are a queue that needs an owner.
More articles