[APPRENTICE] Advanced math or AI alignment. I'm bad at getting homework done and good at grokking things quickly, so the day-to-day should look like pair programming or tutoring.
[MENTOR] See https://www.lesswrong.com/posts/MHqwi8kzwaWD8wEQc/would-you-like-me-to-debug-your-math. The first session has the highest leverage, but if my calendar doesn't end up booked (there is one slot in the next two weeks booked out of like 50), more time per person makes sense. My specialization is pattern-matching to correctly predict where a piece of math is going if it's good. When you science that art you get applied category theory.
[NORMAL]/[meta] Apprenticeships typically start with training but this is in exchange for continued labor from the apprentice for some period after they achieve a useful level of competency. I don't think this part has been made explicit, which means the mentors and apprentices may have different expectations. Did you imagine that this would be training only?
Training can be a significant time cost to the mentor. It's only fair that they be compensated somehow. (The apprentice is compensated with new skills.) If the mentors are simply paid with money as they go, this amounts to tutoring rather than an apprenticeship. If the compensation is some period of labor, then the amount of time and nature of the work should be agreed upon in advance, perhaps after a brief tryout period. Also, what mechanism do we have to punish defectors? The apprentice may simply cut off contact once they've achieved competency, refusing to perform any more labor. For standardized apprenticeships in a trade, they could be taken to court for breach of contract. For apprenticeships over the Internet where we're not even certain of each other's real names, this seems harder. I'm open to ideas.
Sometimes the best way to learn something is to teach it. The mentors can hone their own skills by teaching them, so mentoring an apprentice who defects is not a complete waste of time, but it's still not clear if that would be worth it. Perhaps a useful type of compensatory labor could be a period of co-learning where the apprentice helps the mentor study their skill more deeply.
Something like a mutual mentorship seems more fair. It would be an exchange where both parties teach the other a new skill. This would work really well if there's a match, but what are the chances that you know the skill your mentor wants?
An alternative system, which might work better for us, is a gift economy. The exact details can vary, but in a gift culture, you attain status by being generous, and lose status by being in someone's social debt. If you're indebted to a mentor, you can regain status by paying it forward: mentoring someone else in the community. For this to work though, we'd have to be open and public about what is happening, to bring the social pressure of the group into play.
Even this could be gamed to some degree, using sock puppets, but doing so would be somewhat costly anyway, especially if the apprentices have high karma. Maintaining the puppet might not cost quite as much as maintaining an apprentice, but it's still a cost. In this sense, your LessWrong karma can be thought of as a kind of credit rating: how much you are invested in the community, and thus subject to its social pressure. Those with new accounts (say less than 1000 karma?) have lower social costs when they default. They could simply make a new account and start over. Not sure where the cutoff should be.
We should come up with some way to compensate for this. It occurs to me that we could have them post a bond to some escrow, to be paid to the mentor if they default. But then we'd need a trusted third party and a system to judge. Another possibility is some kind of verifiable community service as a prerequisite for low-karma apprentices. Maybe you guys can think of others.
Testimonials could form part of this system. Mentee provides a testimonial that mentor is knowledgeable on the subject, able to communicate that knowledge, able to help someone progress, etc. This is both useful prestige and evidence of valuable skills.
[MENTOR] I expect to be busy for the near future. However, please contact me anyways - my email is on my LW profile page.
[MENTOR] Research on aligning language models. This includes developing strategies for:
[APPRENTICE] I'd be interested in starting to work in AI alignment, and aligning language models sounds particularly interesting.
I'm an incoming new grad SWE at Google and have a degree in Statistics and Data Science, to give an idea of my background and skillset. I have a modest amount of experience with machine/deep learning, mainly through coursework and a few personal projects (taken and TAd a class on Natural Language Processing, taken a survey Deep Learning class, wrote my undergraduate thesis on applying neural networks to causal inference problems). I've read a lot of the Alignment Forum and LessWrong sequences on AI alignment (Iterated Amplification, Value Learning, etc.) and would like to move into AI capability and safety research at some point in my career.
I may not have the right background or enough experience to be a useful apprentice for you but if you'd like to get in touch, you can reach me at anshuman.radhakrishnan@gmail.com
[Apprentice] I'm interested in this, may be over committed at the moment, but would really enjoy an intro chat.
[MENTOR] I just finished high school last year so my primary intended audience are probably people who are still in high school. Reach out if you're interested in any of these:
[APPRENTICE] Navigating college effectively (deciding what to aim for and how to balance time commitments while wasting as little time as possible). I don't know how much I should care about grades, which courses I should take, or how much I should follow the default path for someone in college. I'm aiming to maximize my positive impact on the long-term future. A message or short call with someone who has (mostly) finished college would be great!
email in bio
Hey, I don't feel qualified to mentor someone on navigating college effectively, but while in college the best advice I read by far (even though it wasn't specifically targeted at college students) was Nate Soares' Replacing Guilt sequence :)
I wish I had someone like you to mentor me back when I was in high school. Those topics are so cool.
[MENTOR] Parenting. Pragmatics of real-life parenting (four boys, the oldest now 17, the youngest 9) from crying in the night, changing diapers to troubles at school, and conflicts of interest. Also a lot of ideas about what goes on in their brains while they develop. We write a weekly development diary and over time you spot patterns. If you want to know my style you may have a look at some of the parenting posts I did here (and the comments in the other posts in that category).
[APPRENTICE] My first is due in November. I've had a very hard time finding evidence-based parenting resources on the Internet that aren't for extremely bad situations like poverty or abuse. I feel a burning need to be able to roughly model this kid's subjective experience on a rolling basis because I suspect that's what will make me the most emotionally effective AND let me impart the most rationality-adjacent thought habits. But the books I've come across have been either 1) "it's all Piaget!" which seems somewhat outdated or 2) "Piaget is a good framework but outdated, and I've read some studies, but I'm terrible at synthesis!".
Even just a reading list would be super great. Or a list of 10 heuristics for making parenting decisions. I feel like I need some kind of systematic approach.
I saw in one of your parenting posts that you cited parentingscience.com, which I'd come across in my searches and looked promising, but I couldn't get enough clues from the site itself to figure out if it was a good foundation.
I suggest that we have a call so you can figure out whether you generally agree with my model of child development and my approach to parenting. I guess that informs the resources recommended.
Best time to have a call is after 9 PM CEST.
supposedlyfun and I arranged a mentoring and you can find a summary of our collaboration in this LW post.
UPDATE June 27, 2021: I got a lot of replies. My schedule is full and I am no longer taking mentees at the moment. However, I am still open to one-off video calls if you'd like to say hi or you'd like me to help you get started with something.
[NORMAL] What an opportunity! Kudos to Zvi for lowering the activation energy by creating this thread. You win this week's Catalyst Prize.
Here is a link to the Don't Worry About the Vase sister post for those who don't have it.
[APPRENTICE] Quantitative finance. My goal is to eventually start my own hedge fund. I have experience in mathematical modeling and I have hand-coded my own small data machine learning algorithms, but I haven't applied these skills to a financial context yet. I want to learn more about this industry but I don't want to move to New York or commit myself to working 9-5 for someone else.
[MENTOR] Writing. I am good at writing nonfiction. I got linkposted by Scott Alexander. I even helped write advertising copy that got my startup on international news. I'm not willing to be your editor (for free) but I am willing to be my editor. There are many things I want to write but don't have the time to write myself, like a sequence on Adversarial Learning Environments. My offer: I help you learn to write better on the condition you write about my ideas on the topic of my choice and then we both get our names on the post. Prerequisites: [1] Basic competence at writing. [2] The skill of translating ideas into words (legible thinking).
[MENTOR] Programming. I am a self-taught software engineer. Proper guidance could have saved me years. Prerequisites: You must be very smart and self-motivated. You must learn fast and work faster. I expect you can read an entire technical book of this difficulty in a week. What I get: software development. What you get: education, experience and a portfolio.
[MENTOR] Agency. People routinely comment on my agency and their own lack thereof. If you want someone to help guide you through savage trials of pure will then I'm your guy. My expectations of you are extra high on this one. [See below.] If you fail to complete a single assignment on time then we're done. What you get: A life of your own. What I get: This is an experiment. I'm curious to what extent agency can be learned.
[MENTORSHIP EXPECTATIONS] Free mentorship offers tend to attract flakes so if you show up one minute late to a meeting then we're done. If I message you and you do not reply in a reasonable timeframe then we're done. I'm not going to be your therapist. Expect something more along the lines of a drill sergeant.
[CONSPIRACY] I believe we might be in an AI overhang and that AI alignment depends on winning the race to an AGI. Still figuring out the details on this one since it's a jungle of secrets and infohazards. More information is available here.
You can contact me via Less Wrong private message or by sending me an email.
[APPRENTICE] I'd like to take Isusr on what is being offered. Writing directly as well. Following in danohu's good idea in leaving a message to show that Isusr's proposal and the apprentice system is attractive.
[APPRENTICE] I am curious about "Agency", but I would love to know more about what kind of exercises/messages you have in mind. I am passionate about self improvement and growth. I do catch myself occasionally not accepting some rational argument for emotional reasons or not working on the most impactful problem. I would love to improve this. Happy to switch to private messages, but I thought that others might have similar questions.
My proposal isn't about increasing rationality. It's about willpower, self-control and initiative. I propose a four-phase system.
Free mentorship offers tend to attract flakes so if you show up one minute late to a meeting then we're done. If I message you and you do not reply in a reasonable timeframe then we're done. I'm not going to be your therapist. Expect something more along the lines of a drill sergeant.
I like this.
[NORMAL] I don't. johnswentworth's The Apprentice Experiment, which kicked off this topic, specifically referenced Selection Has A Quality Ceiling.
To me, this looks like selection for agency more than training for it. The whole point of doing apprenticeships was to reverse this so we could break through the quality ceiling. We've lost purpose here.
[NORMAL] I agree that I have created a filter for high agency. My intention is to increase agency from high to higher. It seems to me like the way to get the best people is to filter and then train.
The writing offer sounds like exactly what i need. I'm in
[I'll write you directly; also posting here as evidence that The System Works]
[MENTOR] computer programming in Python. My level is pretty advanced now, but I certainly don't have everything 100% memorized. This could be a good refresher for me, but I'm pretty busy right now, so I have limited time for this.
The exact curriculum could depend on my knowledge and your interests, but, for example:
I could probably also help with certain other programming languages, but in some cases this might be co-learning rather than mentor/apprentice.
[APPRENTICE]
I am a total beginner who is interested in learning Python from you. I appreciate that you are willing to try with someone who seems to lack talent, because that is where I presently am. I want to be a programmer, and I have tried several learning programs. But each time I was unable to get very far with them, even when I did everything I could to focus.
My preferred mode of communication is Discord, where I can be found as Roland✩#5579 (yes that is a star). I will accept any incoming friend request so that we might talk. If that is not ideal I can also privately supply skype, email or phone numbers.
[MENTOR kind of]
Non-standard perhaps, but I think good to toss into this pool. The Monastic Academy takes on two-month apprentices (with a free bed and free vegan food). You study meditation, the Buddhadharma, fundraising, grounds work, cooking, cleaning, how to give and receive feedback, and how to be a good community member (we all live and work together). Ideal if you are bought into social collapse / climate collapse scenarios being one of the major challenges the planet faces and at least somewhat bought into the idea that the major leverage point is the mind itself (on every scale).
Location: Lowell, VT
More info: https://www.monasticacademy.com/apprentice/
DM me if you want to ask me stuff.
[MENTOR] Machine learning. I do it for work, but I'm a bit behind the frontier and this would motivate me to catch up. You should already know how to code in Python, and have taken multivariable calculus and linear algebra.
[APPRENTICE], possibly. I'm very comfortable with the Python and minored in mathematics, which included Calc III and Linear Algebra. It's been a while since I've used the calculus though. I felt like those classes were rushed.
I've done backend Python/Django work, but was never all that into web development. I just kind of fell into it because of the demand. Machine learning seems more interesting as a job, but I'm uncertain of this, having never actually done it.
[APPRENTICE]: I am a brand new startup founder with, like, 'a few thousand' in funding. (This achieves a surprising amount in the esports industry but is not remotely enough for what I want to do.)
I am 23 and do not know remotely enough about the world to build a successful business. I am interested in fixing this. I will generally be very grateful to people who teach me things that turn out to be relevant.
[NORMAL] The very basics: https://www.goodreads.com/book/show/9512985-the-personal-mba not sure if that's above or below your level, but I thought I'd throw it out there.
Thanks! I've spent a couple years working as an investor, so I'm reasonably happy with my basic knowledge - but I'll skim this and see if there's anything for me.
[APPRENTICE] for character development / lessons for life. Not sure what exactly I expect, but I want to create workshops about these things eventually, as my contribution to raising the sanity waterline. I still have a lot of learning to do first, so I'm asking for an opportunity to do so.
What do I mean exactly: Basic useful skills like public speaking, charisma, communication (speech & text), financial intelligence, self-, and project management, teamwork-skills, leadership, and whatever else a 'master' would have at least a basic understanding and capability of. I'd be thrilled if you can just take on a single one I mentioned, or anything else I didn't.
[MENTOR] Investing (mostly passive ETF for wealth preservation).
[MENTOR] Django, including deployment on VPS.
[MENTOR] Switching to Dvorak from QWERTY and deciding whether you should do this.
[MENTOR] Using computer in a more sustainable for your body way (wrists, eyes, back).
[MENTOR] Feeling happier (not an expert, but can point to many resources).
[APPRENTICE] I would like to feel happier! And I think I need help with it, and I've tried therapy several times and that did not seem like the right kind of help.
[APPRENTICE] Investing. I think I have made some decent choices but would love to have an outside view from someone who feels confident about what they’re doing.
[APPRENTICE] I'd like to learn more about investing and finance, to the extent that finance will help me understand investing. My goal in this is primarily being able to manage my own money/investments and ultimately to be able to reach financial independence so that I can work on effective/impactful causes and donate more. I've been reading more about cryptoassets lately, but don't have strong opinions yet about where is best to point my attention in this field.
[MENTOR] I'm a self-taught programmer who's done mostly web-programming (lot of Python, Django, some FE, some JS/node). I'm potentially available for some amount of mentoring in that area.
To understand investing, at a high level, read
Curious why people have been downvoting this comment. Thanks for the recs. I already have read several of these and have some of them on my to read list, but some are new.
[APPRENTICE]: I would love to be an apprentice / receive advice about bootstrapping online businesses (non-exclusive example: Software-as-a-Service). I find this much more intrinsically rewarding and interesting than mine current 'conventional' job (especial being able to focus on areas I find important and to be less driven by external incentives). Ideally I would like to boostrap myself into working full time on my side projects (that is completely replace my current source of income).
I am a backend developer learning frontend these days (React + css frameworks). I think I already have technical skills to make and run services. My weakest area is analyzing the market and demand / finding problems to solve. Also I suspect that I might be too conservative and not trying out enough ideas or committing to ideas without seeing strong demand.
I already have users on a free product I made and even received a donation.
[MENTOR] I can mentor on the technical and partly on the organizational side of bootstrapping businesses. I created the technical infrastructure of a fintech startup from scratch (I'm the CTO, we have ~50 employees right now, tripling every year). I have a systematic plan for what to do when in IT and a rough idea in all other areas. I understand that you know how to provision and deploy web services but you might want to have an idea of what else you need and in which order.
About analyzing marked and demand I suggest you have a look at Paul Graham's Twitter. The main suggestions he comes back to being
"Build something you yourself want" is to startups what "Write what you know" is to writing.
https://twitter.com/paulg/status/1101071526395342850
and
When startups think a partnership with some big company will be the silver bullet that makes them grow, they're usually mistaken. What makes you grow is building amazing things and making customers love you.
This is an instance of the more general rule that when startups think anything other than building great things and engaging with users will make them grow, they're usually mistaken.
[APPRENTICE] Working on and thinking about major problems in neurosymbolic AI / AGI. I:
[MENTOR] Anything I mentioned in my academic background in the apprentice section. I should have more time at the start of 2022.
[APPRENTICE] I have a bachelor's degree in Math and I want to get a job in Quantitative Finance. I don't know much about Finance, or the math that is specifically used there, though I am very confident in my math skills. I would like to get a better understanding of what the job is like and also what kinds of skills they are looking for.
Have you considered just doing some BTC/BTC-PERP arbitrage, or betting on politics and sports? You'd probably learn what skills they're looking for, gain some of them, and make money while you're at it...
[APPRENTICE]
Book writing - I'm publishing my first technical book this year (no name publisher, they gave me a good deal) and I want to publish a fiction book next year (this one paid out of pocket) | I'd love help from anyone that's gone through the process and published a few books, especially if your starting point was blogging style articles.
Trading - I eer on the side of Taleb here in thinking that the chances someone sharing true knowledge on the subject is low. But if anyone would like to help me with automated trading I'd be very happy to listen. My main focus is using a mix of high compte timeseries analysis for the prices plus NLP to analyze odd sources of data, with a focus on instruments popular among amateur traders, where I assume it's easier to find inefficiencies (but maybe I'm entirely wrong, that's why I'd appreciate advice)
[MENTOR]
Remote working - if you need advice on how to find remote work, how to manage the resulting cost benefit matrix (e.g. more control over taxes, higher freedom to travel) I'd be willing to help depending on your starting point. Background/Credentials: Remote work and travel for 3 years with a low-effort and very legible 9% tax (real, including VAT, social security and such) setup
Getting in tech as a dropout - If you're in the 17-20 age range and wish to skip the college racket and instead just work towards a low six figure job in tech. I did this, helped a few people along in doing this. But it's certainly more cost effective and simpler if you're younger and haven't invested in a diploma mill yet. Background/Credentials: I did this starting age 19, currently working in ML research, but I was mainly working in specialized programming and small team management roles before.
Maybe self hosting and refactoring if anyone here is interested. But kinda niche areas.
[APPRENTICE] Interested in your offer about getting in tech without college. I'm 20, I've learned python on my own, and I'm currently looking for a job without much success. FYI, I am from Israel, so I don't know if it's as relevant.
[APPRENTICE] Writing good code. I have been developing on Ruby on Rails for the last five years with zero formal training. My website backend has a booking engine, email parsing, a dashboard pulling info from many APIs and can modify info on other websites without APIs (e.g. airbnb). I can code what I need but since no one has ever looked at my code I’m certain there’s a lot of bad practices in there, in spite of reading up on good practices and correcting as much as I can.
To get feedback on your existing code, try using an autoformatter, a linter, and a type-checker (I can recommend specifics for Python, but not Ruby). For more general feedback, I found that contributing bugfixes and later features to open-source projects that I already used taught me an enormous amount; if you make an effort to respect maintainers and their time they're almost always amazingly helpful and knowledgeable people.
[APPRENTICE] for anything biology-related. Bonus for working in bioinformatics or simulating (metabolic) pathways. Also interested in joining some startup doing work in this area.
I'm doing a computer science master’s degree and visiting introductory biology lectures on the side, I'd love to build up on those. Additionally, I'm getting deeper in ML and simulation, and would be interested in applying it to biological data. I am currently reading a number of biology papers for creating a presentation which is to be held next month.
[NORMAL] Already posted something quite similar as a reply, but posting as a main comment to indicate broader interest as well.
[APPRENTICE] AI Alignment, Machine Learning Research
I'm an incoming new grad SWE at Google and have a degree in Statistics and Data Science, to give an idea of my background and skillset. I have a modest amount of experience with machine/deep learning, mainly through coursework and a few personal projects (taken and TAd a class on Natural Language Processing, taken a survey Deep Learning class, wrote my undergraduate thesis on applying neural networks to causal inference problems). I've read a lot of the Alignment Forum and LessWrong sequences on AI alignment (Iterated Amplification, Value Learning, etc.) and would like to move into AI capability and safety research at some point in my career.
I may not have the right background or enough experience to be a useful apprentice for you but if you'd like to get in touch, you can reach me at anshuman.radhakrishnan@gmail.com
[APPRENTICE] Any AI alignment (or related) stuff. I have an average of maybe 5-10 hours of time per week to give to something over the next ~10 months while I finish a chemistry master's degree. I have decent experience programming (a few languages as part of my degree) and some in pure maths (BMO1/2 which are high school-level national Olympiads in the UK).
I'd be interested in anything that would let me get involved with or get a feel for the field in the way just reading random papers and posting here doesn't.
[APPRENTICE]
I'm seeking improvement in hardware design, electrical engineering, signals processing, control systems and neuroscience (each somewhat sporadically but they're all connected). I learn quickly but I'm at a loss for tangible projects, so if you have any suggestions for any of those topics and/or are willing to commit some time to mentoring me in completing them, please share!
[MENTOR]
Isn't the academic grad school basically this same model, at scale? I do not see any improvements here that are scalable.
[NORMAL]
Scalable? Maybe not. But this model has significant advantages over grad school.
A while back, LessWrong poster Aysajan put up a post asking to be someone’s apprentice. He talked about it with johnswentworth, who I recently confirmed via meeting him in person is awesome and does reliably interesting work, and an apprentice experiment was born.
As John says, you gotta admire the chutzpah. Asking for what one wants is a known to be successful but highly underused strategy, I presume mostly because of the permanent global chutzpah shortage and the associated danger that it might result in mild social awkwardness.
In addition to the highly successful use of chutzpah, this also points out that apprenticeships are also a known to be successful but highly underused strategy. My feelings about so-called ‘schools’ are well known, but education is great, and apprenticeship is one of the best ways to get an actually excellent and useful education.
I’ve been an apprentice, regardless of whether it was called that, and it was awesome. At Jane Street they have a formal education process, but the core of how you get good is an apprenticeship. You learn from the best, by working with the best and asking them questions. I believe I am a natural trader, but I am good largely because I learned from working directly with three of the best in succession at three different jobs. I know Magic and game design by spending tons of time talking and working with top Magic players and game designers.
I’ve been a mentor of sorts a few times. That’s mostly been great too, and I’m plausibly taking on another one now, although I’m cheating because he is already exceptional and largely does not need the help.
Thus, we have two overpowered strategies here that the world needs more of: Apprenticeship and Chutzpah. Lowering the activation energy required for either or both of them seems great, as does providing encouragement.
Both can of course be overused and abused. Too much of the wrong kind of chutzpah is no good for anyone, and apprenticeship can turn into a bunch of not very useful unpaid work, or end up holding people back. In the context of posts like this, I am not much worried about either of these failure modes.
For this post/thread I will focus on apprenticeship. In particular, I want to see if I can give social permission and a coordination mechanism that can perhaps take place in the comments (reminder that my posts have two comments sections, the primary one at DWATV and a secondary one at LessWrong).
Replies to this post should take the form of any of the following:
Cost of speaking up is low, potential upside is high, and hopefully not too much chutzpah is now required.
As for me: Right now, my main focus is on game design. We hired two new designers this week, both of whom I’m super excited to work with on Emergents, so for now my card is full. Depending on how much bandwidth I prove to have, I could consider mentoring someone at some point in either game design (on either Emergents or my other project), in trading (although I kind of already know who that would likely be if it happened), or someone I trusted sufficiently who wanted to work on Aikido. I might also want to be an apprentice at some point, likely in something AI-alignment related, but that would be a long way off.
I’ve also got another project potentially in the works that I’d love to work with someone on, which would involve someone taking on a likely-more-than-full-time job and a lot of responsibility, starting and running a business I want to exist. It would be hard work and require a self-starter, and all that, but won’t require you to do startup fundraising – it would be a real business, and would succeed or fail organically. It would be local to New York City. If you think you might be the person for the job, you can mention that with [NYCBUSINESS], tell me as much or as little as you’d like, and if/when I’m ready I will reach out. Again, this commits you to absolutely nothing.