| CARVIEW |
Don’t Worry About the Vase has many words addressing many things.
Primarily it is now a blog about AI.
Each week there is an extensive AI update on a wide variety of topics. I start out by covering language models offering (or not offering) mundane utility, move on to questions of mundane harms and benefits generally, and see what new things have dropped on us since last week, including new deals, products and initiatives.
I also cover issues surrounding catastrophic and existential risk from AI, and the discourse around those issues. I hope you also stay for that, I think it is important, but everyone is encouraged to read the parts they find most valuable. My suggestions for what is most unusually important will be in the table of contents in bold.
There are usually five posts per week, including the AI update.
I also cover a variety of other topics with frequent roundups and occasional other posts, including housing and traffic, childhood and education, fertility, medical and health issues and almost everything else, including my loves of sports and gaming.
The only exceptions are heavily political topics, where I look for places to ‘pull the rope sideways’ but there are hills one can easily die on that would only distract, bringing more heat than light. So I avoid those as much as possible.
Why Subscribe?
I like to think my posts are worth reading for many people. Subscribe if you think my stuff is worth reading, and emails are how you’d like to read or be reminded there are now posts. If you don’t agree with both those things, then don’t do that.
Subscribing gets you full access to the newsletter and website. Never miss an update. I’m continuing to support both Substack and Wordpress versions for now, so you can read and comment at either location. Substack is primary.
Like Twitter, This Website Is Free. Paid subscriptions are enabled and encouraged, but they’re not functional, and content is not gated. At some point I may need to make a comment thread subscriber-only or something, or maybe make a subscriber-only open thread, but I have no plans to do even those things at this time.
Paid subscriptions give me money, which I’m grateful for, and which is motivating, helps me justify more time spent writing, and gives me a budget to help with that, including potentially hiring help.
That’s all that they do. Please, please do not subscribe unless giving me money is something you want to do, and it’s something you can afford. If you want to make a larger contribution, the ‘founding member’ option lets you do that, or you can contact me for other options.
Stay Up-To-Date
You won’t have to worry about missing anything. Every new edition of the newsletter goes directly to your inbox. However, due to the nature of weekly updates and other speed premium, corrections after posting are common, so if you want to get the most accurate version it may be good to click through to the website version.
How to Contact Me
If you want to reach out privately, your best bet is to DM me on Twitter (I’m TheZvi), PM me on LessWrong (I’m Zvi) or send an email (which Substack lets you do directly).
For those who have my number, Signal, WhatsApp and text messaging work well.
Substack messaging frequently hides links entirely, and is otherwise not a great means of communication. I will eventually see your message, but this is deprecated.
How To Commission or Hire Me
If you only want a few minutes of my time, or simply want to share something, by all means reach out to do that.
If you’d like more than a few minutes of my time, I am available as a consultant, either online or here in New York City. If I have to travel, that will be a lot extra and depend on details.
If you want to commission or hire me and don’t have other contact info, you can email me through Substack, or you can DM me on Twitter or PM me on LessWrong.
My fee is $1,000 per hour, with a two hour minimum. I don’t waste time. I won’t do net negative jobs.
My discount fee of $500 per hour, with a one hour minimum, is available for clearly positive sum fun non-commercial projects, including advice for those looking to work on such projects, or for work on games.
You can hire me to write a post using that time, which depending on details can involve either fee. Different types of posts take radically different amounts of time. If I already know what I want to say, I can generally write one in three hours. Effort posts, especially involving research, take longer.
I occasionally can be persuaded to come to conferences and workshops and such, or to come give a talk, if they seem sufficiently high value, but I cannot do this that often, so I will probably turn you down unless you are hiring me.
Journalistic Principles
At this point I consider myself a journalist. I operate on more source-friendly rules than the typical journalist.
If you tell me something is fully off the record, or anonymous, or on background, I will fully honor that request. I take confidentiality very seriously. This includes glomarization, as in systematically acting so as not to give away confidential info via what things you choose not to say.
This is not standard journalistic practice, but I allow this to be invoked retroactively, after the statement is made, although what has already been said or posted cannot be taken back and I won’t retroactively censor posts in this way.
With other journalists, remember you have to tell them this in advance.
If I think you meant or should have meant for something to be confidential, I won’t laugh at your stupidity, instead I will point this out and ask you about it.
In general, I strive to not make my source sad about the decision process that led them to talk to me, unless Slussian Protocol is involved (as in, they knew what they signed up for and it was clearly fair game and they did it anyway).
I also strive for accuracy, especially in how others are quoted, including paraphrases, but also in how I state the views of others, as well as everything else. If you are in any way unhappy with what I have said, and feel it to be in error, please reach out to me via email, Twitter DM, LessWrong PM or otherwise, and I will do my best to fix the problem in the online version. If it is important to you, and I discover an error, I can also issue a correction in a future post, for those who read the emails.
If for any reason you are generally unhappy, and do not want to be part of the ‘Zvi Mowshowitz Substack Universe’ in general and instead would like me to ‘keep your name out of my mouth,’ then I will try to honor that request as much as I can. Obviously if you are sufficiently entwined with events this may not be possible.
Technical Notes
Quoted text is sometimes edited for obvious typos and to fix capitalization, or condensed for space (in which case I will use … to indicate a gap).
I will also insert or delete paragraph breaks, and occasionally use [brackets] to indicate a paraphrase designed to correct for a shift in context.
I do my best to edit quotes for length and readability while preserving all relevant context. This includes formatting things as numbered lists.
These changes will typically be unmarked.
When in doubt and where it matters, do check the original source.
I do my best to indicate what level of doubts I have about the credibility of the sourcing for each note. If something seems sufficiently iffy I will either be very explicit on that or hold it back.
However, with speed premium and the cost of extensive verification, and the dedication to keeping access fully free and open, I cannot do full fact checking on everything. There will be errors, either factual or logical, from time to time, and I am not sure if I currently commit too many or too few. When that happens, please mention it in the comments, and I will do my best to address this and fix it quickly.
If you wish to quote or republish this material on the internet, you may do so freely under the condition that (1) you clearly label Zvi Mowshowitz as the author and (2) you provide a link back to the original source. If you do that, go nuts, reprint at will.
However, if your website typically compensates guest authors for posts, and you reprint a post’s worth of stuff, you agree to compensate me in line with your typical compensation for that much material.
You may use this material to train your AI, if you agree to compensate me in line with any other deals you make for compensation to authors or providers of other data. I reserve the right to demand compensation notwithstanding the permissions granted above.
If you train an AI to mimic me and my style and thinking in particular, that is pretty cool, we can work something out, but no you don’t get to make money off that without sharing a cut.
I have signed NDAs over the years, and have agreed to keep various things confidential. If I am asked to keep something confidential, I do it.
I am not under any NDA or other agreement that would preclude me from revealing that the agreement exists.
