gmays

Yes, much better. ChatGPT/Claude/etc. are useful the times I want extra explanation to help connect the dots, but Math Academy incorporates spaced repetition, interleaving, etc. the way a dedicated tutor would, but in a better structured environment/UI.

Their marketing website leaves a lot to be desired (a perk since they are all math nerds focused on the product), but here are two references on their site that explain their approach:

- https://mathacademy.com/how-it-works

- https://mathacademy.com/pedagogy

They also did a really good interview last week that goes in depth about their process with Dr. Alex Smith (Director of Curriculum) and Justin Skycak (Director of Analytics) from Math Academy: https://chalkandtalkpodcast.podbean.com/e/math-academy-optim...

Good luck! You should check out Math Academy, it's more effective/efficient/cheaper but also a good supplement since it's accredited.

I recently turned 40 myself and I'm working through their Foundations courses (made to help adults catch up) before tackling the Machine Learning and other uni courses.

Have you tried Math Academy? I think the difference is that it's actually made by a team of mathematicians creating the content manually.

Yes, Math Academy is insanely good. It's a lot more intense, focused learning and less edutainment.

I'm an adult and not a kid, but wrote about my experience after 100 days of using it daily here: https://gmays.com/math

The Math Academy team (including the founders) are also active on X/Twitter: https://x.com/_MathAcademy_

And there's a Math Academy community on X here in case you want opinions from other users: https://x.com/i/communities/1833198423593431339

The hard thing is that it's both a bubble and not.

It's a bubble in the respect that the hype around integrating into existing companies/software is likely often falling flat.

It may not be a bubble in that all of the best/useful/valuable use cases of AI are in new software, which have yet to prove themselves in the enterprise. This makes sense because you can't just bolt it onto existing software/organizations and expect it to work because they're built around the way things used to be, similar how when factories first tried to integrate electricity.

For example, I'm sure Palantir is doing some good stuff, but I just have doubts about how useful AI can be in the context of existing companies. And their valuation seems insane, which screams bubble, especially since they're an older companies and less 'AI-native' than the newer ones, like the clunky ways Salesforce and Microsoft implement AI.

But do I expect startups to continue to emerge that approach problems in AI-native ways that help companies reorganize? Yes, it's just a question about how long it takes these companies to work their way into the enterprise and earn enough credibility to drive organizational change and restructuring.

The 'bubble' question is really about whether this latent/potential productivity will be enough to inflate the bubble before it bursts.

My money is on yes, but rather than picking a winner at the app layer and trying to win the lottery I'm heavily invested in the boring stuff like chips (NVDA) and those building data centers with low P/Es (back when I bought them), thus a lot of room to grow even conservatively.

Comparing AI to crypto doesn't really work due to the utility of AI. If you believe that there haven't been meaningful use cases from the recent generative AI surge, then you might be out of touch.

On the investment side, it's hard to say that since ROIC is still generally up and to the right. As long as that continues, so will investment.

Then biggest gap I see is expected if you look at past trends like mobile and the internet: In the first wave of new tech there's a lot of trying to do the old things in the new way, which often fails or gives incremental improvements at best.

This is why the 'new' companies seem to be doing the best. I've been shocked at so many new AI startups generating millions in revenue so quickly (billions with OpenAI, but that's a special case). It's because they're not shackled to past products, business models, etc.

However, there are plenty of enterprise companies trying to integrate AI into existing workflows and failing miserably. Just like when they tried to retrofit factories with electricity. It's not just plug and play in most cases, you need new workflows, etc. That will take years and there will be plenty more failures.

The level of investment is staggering though, and might we see a crash at some point? Maybe, but likely not for a while since there's still so much white space. The hardest thing with new technologies like this is not to confuse the limits of our imagination with the limits of reality (and that goes both ways).

Awesome improvements. How does this compare to Braintrust? I've played with it a bit and we're gearing up to implement a solution in during the Christmas lull.

We use various LLMs as a core part of our app but I'm looking for ways to more quickly iterate on our prompts, test different LLM outputs against each other, etc. ideally while minimizing deploys. Would Langfuse serve that purpose?

I guess it's individual. My wife flew F-18's for the Marine Corps and is chronically late for everything! So much for on time, on target.

Real estate can be speculative as well. A better way to state that might be "things outside of productive assets and services" are increasingly speculative.

The bright side of this that has me excited is what seems to be a growing sense of optimism about the future, driven by AI entering the public conscious, Space X's "rocket catch" etc. There seems to be a growing belief that the future can and will be better, more so than a decade ago.

I've been using Math Academy daily for over a year now and have been similarly impressed with how much I'm learning with it.

Congrats on your pace, 9k XP in 9 weeks is impressive!

+1 for Math Academy. I’ve been using it daily for over a year now (started October 2023). I summarized my experiences after 100 days here in case it helps anyone: https://gmays.com/math

Good analysis. One correction though, EmailOctopus does offer auto-plan downgrades. Screenshot of the billing page on our account: https://share.cleanshot.com/VJdQPrjP

After trying a few also we ended up with EmailOctopus because of simplicity (we only send plain text emails) and cost. The trick was using their Connect [1] plans so it could send via our AWS account, which is cheaper (we pay $30/mo for the 10,0000 subscriber plan).

I also tried Loops and wanted to love it since they're perfect for SaaS companies, but back when I tried them we just got a ton of spam subscribers since they didn't have any built-in mitigation, so our list (and cost) grew.

But that was in their very early days, so I assume they've resolved it by now and I'd like to try them again at some point since they're much more modern and purpose-built for SaaS (and a YC company).

[1] https://help.emailoctopus.com/article/161-what-is-emailoctop...

"In times of great uncertainty, the relative value of "playing it safe" is reduced, since - for better or for worse - no option can now reduce risk to truly safe levels. And so, paradoxically, in times of risk and uncertainty, it can actually become more rational to think and act more boldly - or more precisely, to bring one's personal risk tolerance to match the amount of external risk present in the system."

I've had a similar experience. I've now done math with Math Academy for 349 days in a row as of today. I'm not going as fast as I'd like due to other higher priorities like my kids and my startup, but Math Academy helps me make the most of the time I do have. I highly recommend it.

I also documented my experiences when I hit the 100 day streak mark here: gmays.com/math

It’s been great for me. As of today I crossed 324 days of using it straight. I wrote about my experience here: https://gmays.com/math

And yes, I’m a total fanboy. I’ve also known the founder for over a decade and he’s been working on it for most of that time. Math Academy came out of him helping his son learn math even before that.

I haven’t found anything close to teaching math than this. It’s legit.

I feel like sending this to all my vegan friends.."Uh oh, what are you guys going to eat now?"

These days a Nasdaq ETF is likely the best bet (it's what I use in my IRA) since it has been conistently outperforming the S&P 500.

I'd be cautious with the more exotic leveraged ETF. Instead, for my non-investment accounts I've been mostly holding the Mag 7 since last year (after rotating out of pandemic stocks following the rebound). I shared my results over the last couple of years on by blog where I document my thinking for future reference/improvement (https://gmays.com/2-year-follow-up-on-buying-the-dip-on-pand...). My current IRR from that portfolio is 83.70% over the last couple years for context.

It's just the natural evolution of tech towards higher levels of abstraction. In the beginning most dev was on CUDA because the models had to be built and trained.

But since there are plenty of more advanced models now, the next level is getting built out as more developers start building applications that use the models (e.g. apps using GPT's API).

So where 5 years ago most AI dev was on CUDA, now most is on the LLMs that were built with CUDA to build applications.

TL;DR:

1) In 2023, 72% of Americans reported that they were “at least doing okay financially”, a marginal decrease from 73% in 2022, but notably below the recent peak of 78% in 2021.

2) Monthly incomes in 2023 saw an uptick for 34% of respondents compared to 2022, a trend that has continued to increase over the last four consecutive years.

3) Among retired individuals who constitute more than a quarter of the U.S. population, 80% said they were doing at least okay financially, surpassing the national average.

I'd avoid straps or belts until you're at least around ~400lb so you develop your grip and core strength. If you think you need a belt, you could try lifting with a trap (hex bar). That's what I switched to recently and it's more comfortable as someone tall and turning 40.

Not an expert, but can give it a shot:

(1) Much development is already moving from CUDA to the LLM, so less of an issue. Nvidia is also doing more work to increase interoperability. Could be an issue I guess, but doesn't seem like it since there's nothing close to CUDA or the ecosystem.

(2) AMD has attracted significant investment looking at appreciation in its market cap, with a PE ration 3X Nvidia's. However, AMD is so far behind in so many ways, I don't believe it is an investment problem, but structural. Nvidia has just been preparing for this for so long it has a temendous head start, not to mention being more focused on this. Remember AMD also competes with Intel, etc.

(3) Hyperscalers already are building their own chips. It seems even Apple used its own chips for Apple Intelligence. It's relatively (which is doing a lot of lifting in this sentence because it's all HARD) not too hard to make custom chips for AI. The hard (near impossible) thing is making the cutting edge chips. And the cutting edge chips are what the OpenAIs of the world demand for training, but releasing the newest best model 1-3 months ahead of a competitors is worth so much.

If anything, I'd say the biggest threat to Nvidia in the next 1-3 years is an issue with TSMC or some new paradigm that makes Nvidia's approach suboptimal.

Coincidentally, a few days ago my wife was giving talks to students in NYC and mentioned how developed the girls seemed--much more than when she was in high school.

> The code isn't that complicated, you could probably implement training and inference for a single model architecture, from scratch, on a single kind of GPU, with reasonable performance, as an individual with a background in programming and who still remembers their calculus and linear algebra, with a year or so of self study.

Great overview. One gap I've been working on (daily) since October is the math working towards MA's Mathematics for Machine Learning course (https://mathacademy.com/courses/mathematics-for-machine-lear...).

I wrote about my progress (http://gmays.com/math) if anyone else is interested in a similar path. I recently crossed 200 days of doing math daily (at least a lesson a day). It's definitely taking longer than I want, but I also have limited time (young kids + startup + investing).

The 'year of self study' definitely depends on where you're starting from and how much time you have, but it's very doable if you can dedicate an hour or two a day.

You might want to take a look at Math Academy. I'm similar to you and have been using it since last year.

My tool to get me through slogs is streaks, so I commited to doing a lesson (or at least part of a lesson) every day and I'm at 198 days so far.

I wrote this at 100 days in case it's helpful: http://gmays.com/math I'm not sure if I'll have time to write an update for 200 days, but maybe at the 1 year mark.

Congrats! Matt and the team at Automattic are a good group of folks. It seems like a great fit, and I'm excited to see what you guys do together.

Yes, from the post:

> Automattic is doubling down on chat after their acquisition last year of Texts.com, a messaging app with a similar mission. Our teams and products will merge, and I will take on the role leading the team as Head of Messaging.

Yeah, I think Nvidia drives around half as much of TSMC's revenue as Apple does...for now.

With iPhone demand decreasing as the purchase cycle elongates (Vision Pro still low volume) and AI demand surges, it'll be interesting to see how much this gap closes. Other Nvidia competitors will certainly emerge over time, which also incentivizes Nvidia to take up as much quota as they can.

TSMC has worked to convert some older fabs to open up more capacity, so I'll be curious to see if that and other improvements yield any additional capacity, given the incentives.

One interesting question is why TSMC isn't taking more profit. I believe it's because Jensen (Nvidia CEO) is quite shrewd and doing his best to ensure there's always a backup plan. Intel is also doing its best at working towards competing with TSMC here (with the support of the US govt), but TSMC's primary advantage seems to be cultural, which is hard to replicate.

The core difference is that the internet was about distribution while AI is about intelligence, which includes the ability to create things.

The internet is probably the closest the comparison, which is why I get your point, but far off in terms of the scale of the implications.

The hardest thing about AI is that we don't have intuitions for it. It will make the exponential growth that the internet brought look linear by comparison.