Bias Filter: Seeing the Hidden Shape of Sentiment
- Troy Lowndes
- Sep 3
- 2 min read
Why we built it
Bias isn’t always loud. Sometimes it’s in a headline that leans, a word that frames, or the subtle way a narrative keeps certain voices at the centre while others fade into the margins.
We noticed this everywhere… newsfeeds, academic abstracts, even daily emails. And while fact-checking tells us what’s true or false, it doesn’t tell us how language is shaping the story we absorb. That gap was the spark for Bias Filter.
It grew out of our work with Spectral Binary™, a framework we designed to capture tone and emotional resonance in language. Where traditional sentiment analysis stops at “positive/negative”, Spectral Binary treats words like waveforms… mapped along a 0.00–1.00 spectrum that captures subtle shifts, intensities, and frequencies of bias.
What inspired it
Two threads collided:
Media trust crisis - In an age of “fake news” accusations and algorithmic feeds, audiences feel like truth itself is bending. But often, what’s bending isn’t just facts… it’s the emotional weight around them.
Human nuance - As creators of Spectral Binary, we saw firsthand how much meaning is lost when language gets flattened. Emotional shading, rhetorical tilt, implied framing… these are as powerful as the facts themselves.
Bias Filter was born to make these invisible architectures visible. Not to assign blame, but to give readers clarity.
How it works
Bias Filter is both simple and deep:
Input - Paste text, drop in a link, or upload a screenshot. The app parses the language… or extracts it from images… and prepares it for analysis.
Spectral Mapping - Using the Spectral Binary framework, Bias Filter scans for tonal signals… where words lean, intensify, or minimise. Every phrase is treated as a frequency cluster, not just a string of letters.
Bias Spectrum - Instead of a single “bias score”, the app produces a spectrum: who’s being centred, who’s being marginalised, and what tonal strategies are driving the narrative.
Comparisons - With News Scan, you can drop in a topic or link and see how multiple outlets frame the same story side by side. Bias stops being abstract… you can see it.
APIs & Integrations - For developers, Bias Filter offers an API (currently in beta) to integrate this bias-mapping into existing platforms, research tools, or dashboards.
Why it matters
Bias is inevitable… we all frame stories. The question isn’t whether bias exists, but whether we can see it clearly enough to make informed choices.
Bias Filter doesn’t pretend to erase bias. Instead, it reveals the pattern. It hands the agency back to the reader, the researcher, the developer: Here’s how this text leans. Now, how do you want to engage with it?
Try it yourself
Bias Filter is live in beta. Every scan strengthens the model and expands its clarity.