loss·less /ˈlôsləs/ adj. audio engineering. Of compression: removing nothing. No quality stripped, no data discarded. The original, preserved exactly.
Audio engineers prove two recordings are identical with a null test: invert one, sum them together. If the result is silence, nothing was lost. This is that test, run on you, before and after using our software.
Silence. Nothing taken. Nothing lost. The signal is drawn as steps because that's how lossless audio actually stores you: discrete samples, held exactly. And the two traces above are literally the same SVG path, reused. View source. Identical by construction is the only kind of identical we build.
Now scroll, and run the same test on everything already installed on your phone. ↓
Modern software is built on extraction. You give something (data, attention, behavioral patterns), and the company monetizes what it takes. The trade is so normalized that most people no longer notice it happening.
the tap. it's in the architecture, not the marketing.
None of these are accidents, and none are bugs. Extraction is a system, and systems have a start date. ↓
“Everything lives in the cloud” feels like a law of nature. It isn’t. It was built, decision by decision, over five decades, and each decision has a date, a name, and a paper trail. The abbreviated log:
They say this is necessary. That useful software requires collecting data. That privacy and functionality are a tradeoff.
That is a lie. It is an architecture choice, not a law of physics.
You can build an expense tracker that never sees your expenses. You can build location sharing where the server cannot read the locations. The question was never can you. It was will you, because extraction is more profitable, easier to build, and already dominant.
Lossless Protocols is the answer to “will you?”
The system must be architecturally incapable of accessing your data.
Not “we choose not to.” Not “our privacy policy says we won’t.” Not “we encrypt it and promise not to look.” The server cannot decrypt what it relays. The database cannot read what it stores. The company cannot extract what it never receives.
This is the difference between a policy and a protocol.
Policies are promises.
Protocols are structure.
A principle is cheap until it's written down as requirements someone can hold you to. Here are ours. ↓
Privacy is the architecture, not a feature. The system MUST be structurally blind to user content. If a feature requires the company to see user data, the feature is redesigned or cut. There is no exception process.
Users MUST NOT be required to trust the developer. Every claim MUST be verifiable: open-source code, published whitepapers, third-party audits. The architecture is the proof. If the principle were ever broken, the evidence would be public and permanent.
Revenue MUST NOT derive from user data, attention, or behavior. Products are sustained by direct value exchange: free core utility, paid extended features, one-time purchases, community support. If a business model requires extraction, it is not a Lossless product.
No system is perfectly private. The irreducible exposure, the noise floor, MUST be disclosed in plain language before the user commits. Radical honesty about the residue is part of the protocol.
The protocol applies domain by domain (finance, location, health, communication, identity), wherever data exploitation does the most harm. Each domain gets its own product, built for that domain’s specific privacy problem, under the same principle.
The key words MUST and MUST NOT are used as in RFC 2119, because “Protocols” in our name is not a metaphor.
Specifications are promises too, until something ships against them and passes. ↓
One theorem, you = you, proven domain by domain, wherever the industry claims it can't be done.
Finance tracking with zero servers, zero accounts, zero data collection. Everything stays on your phone. The first proof that the lossless principle works.
the signal path ends on your phone. there is no wire to tap.
The second proof, applied to a harder problem: family location sharing that requires a server, so the server is built blind. End-to-end encrypted. The relay forwards packets it cannot read.
what we store: ciphertext. what we can read: nothing.
LP-03 … unallocated. Health, communication, identity: one domain at a time, each proving the extractive model was a choice, not a necessity.
Every proof above has limits. Most companies wait for journalists to find them. We publish ours. ↓
In audio, even silence has a noise floor, a residue no equipment can remove. Privacy has one too. Companies that claim total privacy are lying to you the same way companies that claim extraction is necessary are. Here is ours, in plain language.
even silence, measured closely, has a floor. honest systems say where theirs is.
Limits published. Now the last question a careful reader asks: what stops this company from becoming the thing it replaced? ↓
Most privacy companies die by acquisition: the principle survives until someone buys the company that made the promise. Lossless Protocols is structured so there is no one to make that deal with.
I build these tools alone. That's not a limitation I apologize for: it's the guarantee. There is no growth team pressuring me to add analytics, no investor asking why we aren't monetizing the data, no acquirer waiting to inherit your trust and spend it. There is one person, a principle, and code you can read. If I ever break the principle, the code will show it, and I will have broken it in public, permanently.
Indranil Bhuin, developer, Lossless Protocols
Requirement R2 says you should never have to take our word for anything. That includes marketing. So audit the marketing: