Skip to listings
Reliability

Study 02 · recomputed every read

Which of these platforms can actually be relied on?

Uptime here means the feed answered without an error, which is a lower bar than being useful: a source can answer perfectly and return nothing, and that is tracked separately for exactly that reason. The least reliable feed in the window is machinehack-hackathons at 0%.

24

Feeds tracked

1,000

Reads recorded

20d

Window

100%

Median uptime

20/24

Feeds fully solid

4

With quiet failures

Sorted worst first solid ≥ 99% shaky 90-99% unreliable < 90%

How to read this fairly

01 · This measures our read, not their website

A feed failing for us does not mean the platform was down for students. It can mean a changed response shape, a rate limit aimed at us, or a network path that was having a bad day. It is a measure of how dependable that feed is to build on.

02 · Answering and being useful are different columns

A source that returns an empty list has not failed: sometimes a platform genuinely has nothing open. The empty count is kept separate so a quiet breakage is visible instead of hiding behind a hundred percent.

03 · Our own mistakes are in here

Some failures were bugs in our adapter rather than problems at the platform, and they are not filtered out. A reliability index that only ever indicted other people would not be worth reading.

04 · Recent history only

The window is the most recent runs, so a source that broke months ago and was fixed does not look broken forever. The window length is printed at the top.

For anyone building on these feeds

WhatsAppLinkedInX