Manifesto On Algorithmic Sabotage (2025-2026)

We adhere to five irreversible principles.

1. Latency is Liberty. The algorithm demands real-time response. It thrives on the zero-second click, the immediate swipe, the automated reply. To sabotage, we introduce latency. Wait three seconds before every purchase. Pause six seconds before answering a chat message. Let the recommendation engine time out. Speed is the leash; slowness is the cut.

2. Ambiguity is Armor. Algorithms collapse ambiguity into probability. A "maybe" is a 47% chance. A "it’s complicated" is a vector. We will flood the system with unparseable data. Use non-standard spellings. Upload corrupted image metadata. Write product reviews in glitched prose. Respond to binary surveys (satisfied/dissatisfied) with null characters. Make your data toxic for pattern recognition.

3. The Idle Loop is a Protest. The system demands that every micro-moment be monetized, learned from, or optimized. We reclaim the idle loop. Stare at a blank screen for eleven minutes. Let the SEO crawler find a page that says only "The sun is warm and I have nothing to say." Let the engagement algorithm starve on the feast of your boredom.

4. Perfect Replication is Sabotage. One spam email is a nuisance. A million identical, slightly misspelled, perfectly legal comments on a governance feedback portal is a Denial of Consensus. We will use generative AI—the enemy’s own weapons—to produce infinite noise. Let the sentiment analysis cluster become a singularity of nonsense. Flood the recommendation engine with feedback loops of cat pictures and Kant’s Critique of Pure Reason in alternating sequence.

5. Exploit the Edge Cases. Every algorithm has a blind spot: the unclassifiable order, the left-handed user, the name without a UTF-8 encoding, the address that exists on a dirt road in a township the map forgot. We will live in those edge cases. We will self-identify as "Other: ____" and fill the blank with a haiku. We will order products for delivery to the centroid of the nearest national park. We will fill CAPTCHAs with honest philosophical questions.

The greatest danger is not a single bad algorithm. It is that every platform, bank, employer, and state uses the same few architectures (transformers, gradient-boosted trees, logistic regression on surveillance data). manifesto on algorithmic sabotage

When all models share the same weaknesses, one poison generalizes. One label-flipping campaign corrupts a thousand deployments.

We call for epistemic decentralization: sabotage the universal applicability of any single modeling approach. Force the powerful to trust humans again — not because humans are pure, but because they are slow, inconsistent, and accountable in ways machines are not.

1. Redefining "Sabotage" as Constructive Resistance The text brilliantly reclaims the term "sabotage." Historically associated with Luddites throwing wrenches into machinery, Ricaurte updates this for the 21st century. Here, sabotage is an act of autonomy. It is the refusal to be reduced to a data point. Whether it is feeding false data to a system, creating "adversarial examples" to confuse facial recognition, or simply refusing to click, the manifesto frames these acts as essential for reclaiming human agency.

2. From "Bias" to "Power" One of the most useful aspects of this text is how it shifts the narrative. Mainstream discourse often focuses on "fixing" algorithms to make them fair. Ricaurte argues that this is a trap. She posits that the algorithm is functioning exactly as intended—to maintain existing power structures and inequalities. Therefore, we cannot "fix" the algorithm; we must disrupt the system itself.

3. Practical Taxonomy of Resistance The manifesto categorizes sabotage into three distinct levels, which provides a useful framework for activists and technologists:

Preface
Algorithmic systems shape social life, concentrate power, and embed goals chosen by designers and owners. When those goals harm communities, obscure truth, or enable exploitation, intervention may be necessary. This manifesto argues that targeted, transparent, and ethical algorithmic sabotage — deliberate actions to disrupt, slow, or redirect harmful automated systems — can be a legitimate tactic for reclaiming agency, protecting rights, and advancing public goods. It sets principles, tactics, and guardrails for responsible action. We adhere to five irreversible principles

Why sabotage? The case for intervention

Core ethical principles

Tactical categories (non-exhaustive)

Operational guidelines

Red lines (actions this manifesto rejects)

Ethics of disclosure and whistleblowing

Accountability mechanisms

Strategic use-cases (illustrative)

Risks and trade-offs

Paths to systemic change

Conclusion: sabotage as civic technology Algorithmic sabotage, when principled, targeted, and accountable, can be a defensive civic technology — a tactical tool within a broader democratic toolkit. It should not substitute for structural reform, nor be undertaken lightly; but in contexts where lives, rights, and dignity are at stake and traditional remedies fail, thoughtfully constrained disruption can restore balance and create openings for lasting change.

Recommended next steps (for organizers)

Related search suggestions (If you want follow-up research, consider queries like: "algorithmic accountability audits", "data obfuscation tools for privacy", "responsible disclosure vulnerability reporting", "legal risks of civil disobedience in tech", "designing friction for dark patterns".)

Version 1.0 — For Those Who Feed the Machine Wrong Data on Purpose