urs

Just stumbled upon this thread and wanted to share Richard Hamming's classic talk from '86, "You and Your Research."

Then I realized that the funny part is that PG has already linked to Hamming's talk on his site (http://www.paulgraham.com/hamming.html), and mentioned it on Twitter (https://twitter.com/paulg/status/849300780997890048).

There’s a part in that talk that has always stuck with me: he advises to ask yourself at every Friday evening, "What are the important problems in my field?" Not entirely dissimilar from PG’s take on how the educational system in forcing you to commit prematurely often has you overlook this entirely.

In the vein of "great minds think alike," both of them hammer home the importance of working on what genuinely grabs your interest. PG's advice is to "optimize for interestingness" ; Hamming when he says, "If you do not work on an important problem, it's unlikely you'll do important work."

I got a kick out of how both of them advocate for being flexible in our approach to work — especially given how launching and pivoting after learning from your users has also been the PG advice for the better part of two decades in startup-focused essays. PG's all for switching horses mid-race if a more exciting problem shows up , and Hamming shares the same sentiment, stressing the importance of being ready to seize new opportunities. Today pivoting is just default vernacular in startup world, but also cutting losses and getting that fractal and pushing that to its end is worth it.

Curious how has "optimizing for interestingness" played out in your own work or life? Additionally, curious if there are any good HN stories about pursuing research and “pivoting” in fields that are not searching for product-market/fit for a startup…

(Hamming’s talk has been shared countless times here and this feels like PG’s contribution to a similar idea (https://news.ycombinator.com/item?id=35778036)).

This was neat. It reminded me of a similar story in poker history - Stu Ungar. Stu was a phenomenal player who won the World Series of Poker (WSOP) Main Event three times, in 1980, 1981, and 1997. Like Tosa, Stu had an uncanny ability to read his opponents and seemingly predict their moves.

Similar to Tosa, people questioned Ungar's poker prowess as to whether he was cheating or not (it's a pretty deep conspiracy rabbit hole that you can go down if you'd like). People wondered if he had some secret method for reading his opponents — like whether his blue glasses were letting him read marked cards, but it turned out to be like Tosa, he was a product of pretty deliberate and intense practice.

In both cases, it's a weird example of humans being exceptional at pattern matching in ways you almost can't imagine after deliberate practice.

Ungar is most famous for his WSOP win calling down with a ten high to win.

I think a fascinating aspect to all of this is the other idea in any edge based game where nearly all great players will start on a winning streak and how incredibly difficult it is to discern whether you are lucky or good (there's a small chapter in "The Signal and the Noise" about that specifically with poker players).

Big Teachify crowd here!

I will say, Epihub was really the byproduct of Epigrammar (as Mike highlighted above) and us just using the term “hub” a lot. It was hard to mispronounce, we had the .com, it felt like an homage to our previous work, and we didn’t think about it too much afterward.

I do like Teachify though. It’s a pretty great name; I can’t lie.

I think it could work, we haven’t tried it for a conference, but email me directly (first name at company name.com) so we can make sure you’re setup right.

To answer your question, not only is every single string of text customizable, but so is every icon. So you can rebrand and redesign it however you please and the interface will update.

Yeah the original application borrowed heavily from our first product which predates LiveView so we stuck with what we knew. Our first product actually predates contexts and schemas in Phoenix. We started working on it around the time Phoenix was deprecating models from 1.2 to 1.3.

LiveView is incredible, but I haven’t used it enough to know the shortcomings. From the outside looking in, it does seem nice to work with a view that’s already nicely coupled with the app.

Thanks for the question. So a pretty big difference from the outset is the type of customer we focus on (again, correct me if Learnworlds is different).

Where Learnworlds is focused on creators who want to build courses that are asynchronous. We are focusing on live instruction particularly with businesses that have already been handling live instruction so teams of tutors, trainers, teachers, and coaches.

In many ways Learnworlds is quite similar to Kajabi, Teachable, and a whole host of other great tools for building online courses.

In our case, we started with thinking about existing businesses where a primary concern is team management as the instruction is live. Coordinating live instruction already requires a pretty different software stack from an online course.

In a coaching business of twenty tutors, you have to manage twenty instructor schedules against schedules for your students, figure who is owed what, who you have to bill, and provide space for live online instruction.

The last few months, particularly in New York, have made this coordination problem far worse as these businesses look to move online while trying to keep their branding and identity front-and-center.

—

Now a number of our users have already asked to be able to sell online courses/materials, and we have been experimenting with blending asynchronous online courses with live instruction, so it’s on the roadmap. Right now, however, our focus is uniquely bringing existing businesses with live instruction online.

Thank you, to be honest Abinsthe and Apollo go together incredibly well.

I think coming from Rails and switching out controllers for resolvers made adjusting to building SPAs far easier. I still think building routing on the client side feels strange, but it’s way easier when GraphQL just gives you one endpoint to hit and with Apollo you just query exactly what you need.

Great question. We built the product originally to help us bring a tutoring business online as in-person instruction was impossible.

Again, these were businesses built around live instruction so the only option was video. Zoom was the one place where we opted for an integration as there’s simply no way we could build better video than Zoom given a limited (or unlimited) time frame.

We actually built tooling inside to pick locations for your classes and appointments, but so far they have rarely been used.

Next we had a bunch of tools from our first product in Epihub, but something we learned from talking to teachers and having built tools for instructors was that you really didn’t need too much. Teachers know how to teach so the best tooling isn’t something overly prescriptive, but something in a virtual classroom isomorphic to a real classroom like a virtual whiteboard.

Again, we’re really new, so we’re still learning.

I’m always amazed by the sheer number of products Google has built and sunset. I didn’t actually know about this, but it sent me down a rabbit hole. What kind of work was most popular on Helpouts?

Yeah we're currently limited by Stripe Express, but we'll probably look to expand relatively soon. We've been looking at RazorPay for India or even enabling Stripe Standard accounts across the board. Any particular countries you're looking for?

This made me laugh so thank you, but in all seriousness, we started by working through the lifecycle of a customer from the top and push it to the end. The first part being, how do you meet and qualify a new client, to the last part where you're handling paystubs and invoicing clients.

There are a ton of things I wish we handled better, particularly on-boarding where although it's self-serve, it needs work, but we did figure it's better to launch and learn from more users before building more software.

The second thing was we used the product itself to onboard (vip.epihub.com/reserve), and again once we were users, we kept pushing until it worked well enough for us to train people on how to use our product.

Awesome! Feel free to reach out directly or sign up above with Mike and we'll be happy to onboard you personally.

That’s it right there. I basically have the same thing. Worst comes to worst, Since they are visually so close, when I started, I never forgot what did what.

Something cool to try is to blend your tmux bindings with your vim bindings: https://gist.github.com/mislav/5189704

Mislav Marohnić has a super nice method of moving between tmux splits and vim splits the same way so you don’t really get stuck thinking one split is actually the other.

This would drive me crazy, but after a little hacking, my tmux/vim combo feels so nice and snappy.

I’ve always done:

> Mod+| for vertical split

> Mod+- for horizontal split

I’ve been using it like that for years, but haven’t seen it in too many places like that. It just feels super logical since the symbols look like the split!

There's something amusing to seeing $1B doled out over a Google Spreadsheet while some charities a microfraction of the size will spend thousands of dollars on a custom website.

Spreadsheets become SaaS products and SaaS products become spreadsheets in a never-ending loop.

It'll be fun to see where this goes!

It’s been fascinating watching Satya turn Microsoft into a cloud services company over the last few year and this is another great example again turning an enterprise experience into a consumer one.

I like this part: “The defining quality of an ideologue, whether on the left or the right, is to acquire one's opinions in bulk.”

I think another way to slice this is, if you think of the Overton window shifting along the axis of time, the “accidental moderate” does not shift their opinion by the same factor as the ends of the window shift. The “accidental moderate”, in fact, shifts their opinion independent of the shifts of the window.

I just don’t know 100% if I’d use the term moderate as the it’s not necessarily true that all views will equally weight a view left of center with a view right of center (or vice-versa), and moderate could be perceived as synonymous with “average.”

Additionally, by even defining two types of moderate, there is a sense that the word “moderate” means something already. I don’t know, I feel like there could be a better word, maybe if you think of it as a graph there’s a graph-related term, but it’s not coming to mind!

C++ over Elixir cracks me up. I feel like no matter how hard you try, no one really “””knows””” C++ (given library vastness and all that).