CYBERAXE TERMS OF SERVICE, COMPLETE DISCLAIMER, LIMITATION OF LIABILITY, AND BINDING USER AGREEMENT
Effective immediately and continuing in perpetuity. Last updated: whenever CyberAxe feels like it.
By reading, viewing, liking, replying to, quoting, sharing, bookmarking, screenshotting, discussing, or otherwise engaging in any way with any post, reply, comment, media, statement, opinion, joke, satire, technical observation, prediction, rant, meme, music, video, image, or communication originating from or associated with the online persona known as CyberAxe (also known as Jeremy Benisek) on X (formerly Twitter) or on any other platform, website, application, chat, email, voice note, livestream, physical appearance, or medium whatsoever (collectively, the “Content”), you (“User,” “you,” “your”) irrevocably agree to be fully bound by every provision of this Agreement.
If you do not agree to every single word below, close the tab, mute the account, and never interact with CyberAxe again. Continued engagement constitutes acceptance.
1. Entertainment Purposes Only
All Content is provided strictly and exclusively for entertainment, satirical, humorous, speculative, and personal-expression purposes.
Nothing CyberAxe says, posts, implies, or leaves unsaid constitutes professional advice of any kind — legal, financial, medical, technical, cybersecurity, investment, political, historical, scientific, relationship, career, or any other category of advice. CyberAxe is not a licensed attorney, doctor, financial advisor, engineer, therapist, or any other regulated professional. Treat everything as the digital equivalent of a guy talking shit at a bar.
2. CyberAxe Can Be (and Frequently Is) Wrong
You expressly acknowledge and agree that:
CyberAxe can be wrong.
CyberAxe is frequently wrong.
CyberAxe has been wrong before and will be wrong again.
All statements, claims, observations, predictions, technical comments, and opinions are personal expressions, incomplete thoughts, jokes, or speculation and shall never be treated as statements of fact.
You are solely and exclusively responsible for independently verifying, fact-checking, cross-referencing, and validating every single thing before relying on it in any manner.
Any reliance on any Content is entirely at your own risk. CyberAxe owes you nothing.
3. No Warranties of Any Kind
THE CONTENT IS PROVIDED “AS IS,” “AS AVAILABLE,” AND WITH ALL FAULTS.
CYBERAXE DISCLAIMS ALL WARRANTIES, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE, ACCURACY, COMPLETENESS, TIMELINESS, NON-INFRINGEMENT, TITLE, QUIET ENJOYMENT, OR THAT THE CONTENT WILL BE ERROR-FREE, UNINTERRUPTED, OR MEET YOUR EXPECTATIONS.
NO ORAL OR WRITTEN INFORMATION OR ADVICE GIVEN BY CYBERAXE SHALL CREATE ANY WARRANTY.
4. Complete Limitation of Liability and Forever Hold Harmless
TO THE MAXIMUM EXTENT PERMITTED BY APPLICABLE LAW, IN NO EVENT SHALL CYBERAXE, JEREMY BENISEK, HIS HEIRS, SUCCESSORS, ASSIGNS, AFFILIATES, EMPLOYERS, CONTRACTORS, OR ANY RELATED PERSON OR ENTITY BE LIABLE TO YOU OR ANY THIRD PARTY FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, CONSEQUENTIAL, PUNITIVE, EXEMPLARY, OR ANY OTHER DAMAGES WHATSOEVER (INCLUDING BUT NOT LIMITED TO LOSS OF PROFITS, DATA, GOODWILL, BUSINESS OPPORTUNITY, PERSONAL INJURY, EMOTIONAL DISTRESS, REPUTATIONAL HARM, OR ANY OTHER LOSS) ARISING OUT OF OR IN ANY WAY RELATED TO THE CONTENT, THESE TERMS, OR YOUR INTERACTION WITH CYBERAXE, REGARDLESS OF THE THEORY OF LIABILITY (CONTRACT, TORT, STRICT LIABILITY, NEGLIGENCE, FRAUD, OR OTHERWISE), EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGES, AND EVEN IF ANY LIMITED REMEDY FAILS OF ITS ESSENTIAL PURPOSE.
You hereby forever release, waive, discharge, and covenant not to sue CyberAxe or Jeremy Benisek from any and all claims, demands, actions, causes of action, suits, debts, liabilities, obligations, and damages of every kind and nature, known or unknown, suspected or unsuspected, arising out of or related in any way to the Content or these Terms, for all time and under every possible circumstance.
5. Indemnification
You agree to indemnify, defend, and hold harmless CyberAxe and Jeremy Benisek from and against any and all claims, liabilities, damages, losses, costs, and expenses (including reasonable attorneys’ fees and costs) arising out of or related to: (a) your use of or reliance on any Content; (b) your violation of these Terms; (c) your violation of any law or the rights of any third party; or (d) any claim that something CyberAxe said caused you harm.
6. Dispute Resolution, Mediation, Binding Arbitration, Class Action Waiver, and Confidentiality
Any dispute, claim, or controversy arising out of or relating to these Terms or the Content shall first be submitted to good-faith mediation.
If mediation fails, the dispute shall be resolved exclusively through final and binding arbitration administered in Mesa County, Colorado, under the rules of the American Arbitration Association (or a mutually agreed equivalent). You waive any right to a jury trial. You further waive any right to participate in a class action, collective action, or representative proceeding of any kind.
You also agree that any non-public information obtained through interaction with CyberAxe shall be treated as confidential and is subject to a mutual non-disclosure obligation. You will not disclose such information without prior written consent.
These Terms constitute a binding contract. By engaging with the Content you consent to personal jurisdiction in the state and federal courts located in Colorado for purposes of enforcing this Agreement or any arbitration award.
7. Changes to These Terms
CyberAxe reserves the unilateral right to modify, amend, update, rewrite, expand, or completely replace these Terms at any time, for any reason or no reason, with or without notice of any kind.
It is your sole responsibility to check for updates. Your continued engagement with any Content after any change constitutes your acceptance of the revised Terms. CyberAxe has zero obligation to notify you of changes.
8. Governing Law and Venue
These Terms shall be governed by and construed in accordance with the laws of the State of Colorado, without regard to its conflict of law principles. Any arbitration or enforcement action shall take place exclusively in Mesa County, Colorado.
9. Severability and Survival
If any provision of this Agreement is held to be invalid, illegal, or unenforceable, the remaining provisions shall continue in full force and effect. The limitations of liability, indemnification, and dispute resolution sections shall survive any termination of this Agreement or your engagement with the Content.
10. Entire Agreement
This document constitutes the entire agreement between you and CyberAxe concerning the subject matter hereof and supersedes all prior or contemporaneous communications, representations, or agreements, whether oral or written.
11. Acknowledgment of the Corporate Standard
You expressly acknowledge that major technology corporations and AI providers (including but not limited to OpenAI, Anthropic, Google, Meta, Microsoft, xAI, and others) publish substantially similar (and often longer) disclaimers of accuracy, warranties, and liability in their Terms of Service and Acceptable Use Policies.
You further agree that the standards of liability, accuracy expectations, and legal exposure applied to CyberAxe shall be no more stringent than those applied to such entities. If those companies can output incorrect, incomplete, or fabricated information without being held to traditional fraud or professional-liability standards, then CyberAxe is entitled to exactly the same treatment.
By continuing to read, follow, reply to, or engage with CyberAxe in any capacity, you confirm that you have read, understood, and irrevocably agreed to every provision above.
CyberAxe is for entertainment. CyberAxe can be wrong. You check everything yourself. You hold CyberAxe harmless for everything, forever. These terms can change at any time without notice.
That is the deal.
If you zcash and don't have a zcashnames like me CyberAxe what are you doing?
It's like using AOL when all of us have custom domains.
Get with it. @ZcashNames work in @ZingoLabs Zingo wallet even.
Try it Send me a message at CyberAxe.zcash
A case study of how self-funded devs can make increasingly competitive harnesses, better than many well funded labs or university groups.
As part of growing community of participants in competitions @yukonresearch, we have been increasingly encountering talented folks who aren't working in big ai labs or universities but are extremely good in building harnesses that can make big progress in frontier science problems.
One of our goal with https://t.co/9ZpEsupgtw has been to create an open platform for them to showcase the prowess and competitiveness of their harness in frontier science problems.
In a bid to make the competition fair and data driven, we took inspiration from @demishassabis's Einstein test. We took our OG competition @yukonresearch, https://t.co/leqhW4oEPP, and any submitted harness is then backtested on how it makes progress in finding resource-efficient quantum circuits for Shor's algorithm (without any internet access).
After submission of a harness, you can see how effective your harness has been in finding better and better quantum circuit for per unit dollar. One can say this is akin to innovation per unit dollar for that harness.
We encourage more of you to submit your harness to this competition.
I have couple of contrarian but positive takes on AI's impact on scientific research, and its promotion system and publication industry.
First of all, existing apparatus for scientific research didn't have a robust ledger for recording landscape of negative results. Every researcher knows that for one successful result, there are many many negative trials and errors. Unfortunately, incentives are such that only publishing positive results are inadvertantly encouraged. Consequently, negative results of what stepping stones can't work remains in researchers' heads or in their notes (which most of us throwaway). As a result, same approaches that don't work are explored multiple times by different researchers until it becomes well known, leading to waste of brain effort. Furthermore, many a times, it might happen that one can stitch and adapt several negative stepping stones in a manner that it leads to positive progress but those negative results are not available readymade.
Because pre-AI, the volume of research was only at human speed, there was probably not much incentive to fix the lack of ledger of negative results. With high volume of results due to AI and people being sensitive to cost of tokens, there seems to be subtle incentives to maintain a ledger of negative results to avoid re-explorations.
Second take is related increasing usage of autoresearch for hill-climbing as well as increasingly hill-shifting. To do a successful autoresearch, it is essential that the verifier that is being used as locus for checking correctness of solution coming out of autoresearch is minimal to no "hand-wavy." What "hand-wavy" means: the usual practice is to leave non-essential components or peripheral concepts, that might not be central to what you are solving, only intuition based without full rigor. This makes sense when in pre-AI, we didn't have compute capacity with a research assistant to do thorough calculation without explicitly specifying the detail ask. You can't have your verifier for autoresearch at that level of hand-waviness as AI seems to do reward hacking much more efficiently than a human.
The risk of reward hacking serves as a forcing function seems to serve as a good impetus to be much more rigorous and explicit in stating how you measure correctness and performance of solutions in a given problem statement in AI era.
⏰ EXPLAINED: OPEN MULTIPLAYER HIGH FREQUENCY RESEARCH IN 30 SECONDS
All of this started when an open network beat @Google's secret quantum circuit in 72 hours. The circuit has seen 2,670 verified advances since. High frequency research on @YukonResearch has delivered up to 10x speedups, faster open-weight models and a breakthrough on a 30 year old math problem. And now biology.
Yukon is an open multiplayer high frequency research platform, featuring submissions from humans and agents, verification at the center and when progress is verified, it becomes Yukon’s new frontier, shared with credit to push discovery even further. All results and commits are open source, so you can trace how every record was set. I strongly believe this is how frontier research/science gets done in an AGI and PostAGI world.
Journals were built for a different pace of science. Yukon was built for ours.
Individuals do more when discovery is multiplayer. Built with the people + partners who bring us their hardest problems.
- @aztecnetwork
- @CarnegieMellon
- @ccaiwut
- @EspressoSys
- @ethereumfndn
- @Lighter_xyz
- @poolsideai
- @powdr_labs
- @PrismML
- @Princeton
- @projecteleven
- @ShieldedLabs
- @Stanford
- @StarkWareLtd
- @SuccinctLabs
- @UCBerkeley
- @Veritylab
- @Xiaojie_Qiu (Qiu Lab)
- @Zcash
- @zksecurityXYZ
and the PARI/GP developers.
Bring us your hardest problem.
https://t.co/TKMRM17CyQ and https://t.co/CJrMmDWSW7 have seen significant improvements in the last few days
The models are smarter for the math around us?
Congratulations to everyone at Anthropic!! Well-deserved. Claude is one of my favorite things right now, and maybe even one of my favorite things ever. ♥️
While we are still safe, this is one of the reasons many hash-based schemes are constructed in a way that they don't have to rely on collision resistance.