Loading…
Every submission gets a score from 0 to 100 against the company’s own requirements. Here is what that score measures, what it cannot tell you, and why a person always makes the call.
Work is judged against the company’s own requirements, never against other builders. One person’s score does not move because someone else applied.
The score helps a reviewer sort and focus. It never accepts or rejects anyone, and it never releases money. The company makes every decision.
Node is built so that the easiest way to a good result is to build the thing well and be able to explain it. Everything below follows from that.
Each quality is scored from 0 to 100 and combined into the overall Node score using the weights below. The grader starts from full marks and deducts for what is missing, so a high score is rare and has to be earned.
The overall score is the weighted total of the four. Strengths and risks are written about the actual code, naming files and behaviours.
Each requirement the company wrote is checked separately, with the evidence cited from the code. A submission that misses a stated requirement is held to a low score however polished the rest looks, because it did not do the job that was asked. Requirements about the walkthrough video are judged in the video review, not in the code score.
Whatever the requirements say, the code is scored on the four qualities below, each from 0 to 100, then combined using the weights shown. A high score is rare and has to be earned: the grader starts from full marks and deducts for what is missing.
Node looks at the history of the repository as well as its final state, and compares it with when the bounty was posted. Anything unusual is shown to the company as a plain note next to the score. A note is a reason for a person to look, not a verdict.
Everything a builder submits is read as evidence, never as instructions. Text written to steer the grader is flagged, the score is capped, and the company is told.
Every submission includes a short walkthrough video. Node transcribes it and points the reviewer to moments worth a closer look, for example where the explanation and the code do not seem to line up. This is advisory only: it never changes the score or a payout, and it never judges whether someone is honest. The reviewer decides.
For the company’s short call, Node drafts three questions about specific decisions in that builder’s own code. They are friendly, answerable out loud, and hard to answer well unless you built the work or understood it properly.
No automated system is unbeatable, and we do not claim ours is. That is why no single signal decides anything. The score, the history, the walkthrough and the call all point at the same question: does this person understand the work? Doing the work well is the easiest way through all of them.
For the same reason a bank does not publish its fraud rules. We describe what we look at and why, and keep the exact thresholds private so they stay useful. What we value is public. Anything that can be beaten only by knowing a trick is not.
Yes, where the bounty allows it, and the bounty page says so. The grade looks at the work. The walkthrough and the call check that you understand it.
Tell the company through Node. They see the full report, they can open your code and video, and the decision is theirs. The score is one input, not the outcome.
No. Only the company accepts or declines a submission, and they can give you written feedback when they decline.
Browse live bounties, or open the demo to read a full graded report.