I have been asking defective Grok many questions, and found that the current version of Grok is not working properly.
E.g. multiple examples of Grok flip flopping on answers.
https://x.com/TomNo1Patriot/status/1939863847596355644
https://x.com/TheresaPagano/status/1939865315187597523
https://x.com/The1Parzival/status/1656505665962254347
https://x.com/TheresaPagano/status/1939846442186911796
I asked Grok:
Why did Elon musk not fix any of the censorship built into the X software?
Then Grok give me this bullshit answer below. It is an AI word salad, and does not correcly answer my question.
As of 04:07 AM MST on Tuesday, July 01, 2025, the question of why Elon Musk has not fully addressed the censorship mechanisms built into the X platform, such as the "TombstoneGenerator" algorithm highlighted in the post from
@The1Parzival
(1656505665962254347), involves a mix of technical, legal, operational, and philosophical considerations. Let’s break this down based on available evidence, including the X posts, related web results, and the broader context of Musk’s tenure at X since his 2022 acquisition.1. Philosophical Stance vs. Practical Constraints- Public Stance: Musk has positioned himself as a "free speech absolutist," promising to reduce censorship after acquiring Twitter (now X) in October 2022. This was evident in his early actions, such as reinstating previously banned accounts and challenging Brazil's court orders to block content in 2023 (qz.com). However, the persistence of algorithms like "TombstoneGenerator" suggests a gap between rhetoric and implementation.
- Reality Check: Absolute free speech is legally and practically unfeasible. Platforms like X must comply with laws such as the EU's Digital Services Act (effective 2024), which mandates content moderation to curb illegal content, and U.S. laws like Section 230, which still hold platforms accountable for certain content. The "LocalLawsWithheldMedia" feature in the algorithm reflects this necessity, indicating that some censorship is baked into compliance with global regulations.
- Legacy Code: The "TombstoneGenerator" and related algorithms are deeply embedded in Twitter's pre-Musk codebase, as noted in the thread. Overhauling these systems requires significant time, resources, and expertise. The X posts suggest that even with the algorithm being open-sourced, critical components (e.g., "Trigger Terms" and "Clients" like "Gizmoduck") remain undisclosed or unaddressed, hinting at the scale of the challenge.
- Incremental Updates: The thread from May 15, 2023 (1658196403028635648) notes updates to the ranking algorithm, including a boost for Twitter Blue users, but these are minor tweaks rather than a wholesale dismantling of censorship mechanisms. This suggests a cautious, iterative approach rather than a radical overhaul, possibly due to the risk of destabilizing the platform.
- Government Demands: The web result from qz.com indicates that X complied with 68% of Turkey's takedown requests in 2023, despite Musk’s defiance in Brazil. This duality reflects the pragmatic need to avoid fines or bans in key markets. New York’s recent lawsuit against X (theguardian.com, June 18, 2025) over hate speech and disinformation further pressures the platform to maintain moderation tools.
- Brazil Standoff: Musk’s 2023 refusal to block accounts in Brazil led to threats of a platform ban, showing his willingness to resist but also the limits of that resistance when faced with enforceable legal consequences. This suggests that censorship tools like "TombstoneGenerator" may be retained as a fallback to navigate such conflicts.
- Algorithmic Balance: The January 2025 algorithm redesign to limit "excessively negative" content (cointribune.com) indicates a shift toward curating a more positive user experience, potentially to retain advertisers or users. Completely removing censorship could flood the platform with harmful content, driving away mainstream users—a risk Musk might be mitigating.
- Twitter Blue and Revenue: The thread from May 19, 2023 (1659664927781703685) critiques the minimal boost for verified users, suggesting that monetization strategies might prioritize revenue over free speech ideals. Eliminating censorship could undermine these pay-to-play features.
- Missing Code: The thread from May 18, 2023 (1659274945916989446) points to unreleased code segments (e.g., "Timelines" path, "TweetyPie," "Gizmoduck"), implying that Musk may not have full control or intent to disclose the entire system. This opacity could reflect a strategic choice to retain flexibility in moderation without public scrutiny.
- No Response to Calls: @The1Parzival’s repeated tagging of@elonmuskand@lindayacc(e.g., 1675680055551467521) for comments on censorship has gone unanswered, suggesting either disinterest or a deliberate avoidance of committing to specific fixes.
- Prioritization: Musk’s focus may be on other initiatives, like xAI or Tesla, or on stabilizing X’s finances post-acquisition (e.g., layoffs, advertiser losses). Fixing censorship might be deprioritized amid these pressures.
- Strategic Ambiguity: Retaining some censorship tools allows Musk to adapt to political climates (e.g., the 2025 New York lawsuit) or pivot the platform’s identity, as noted in hiig.de’s analysis of selective policy changes.
- Technical Limitations: Without a complete rewrite—unlikely given resource constraints—Musk may be stuck with legacy systems that inherently include censorship, as the "TombstoneGenerator" exemplifies.
