Ali Spittel: Yeah, I think it’s unreal how fast this has turned. The economy in general has clearly shifted with the response to inflation, and then also the shift in the markets as well… I think that a lot of software engineers, a lot of our compensation is based on the company’s stock, and so that has dramatically decreased for a lot of people over the past few months as well… So even if you’re at your current job, you’re probably still affected, even if you haven’t been laid off, so I think it’s a really interesting economic moment of “Will this be a recession?”, which is a very scary thought, I think, for a lot of us… Or is this just a momentary thing? Is it just that things are shifting and we’re kind of correcting for the fact that a lot of these companies boomed really fast due to the changes in our lives, due to the pandemic.

[44:24] I think that we’ll have to see, I don’t think that anybody knows… People who work in finance have predictions, but I think it’s a really hard thing to predict, and there’s lots of conflicting predictions out there… But it’s a tough time, and it’s really wild how fast it shifted from, like, you’ve gotta leave your job, you’ve gotta get a new job while everything’s hot… Software engineers are getting paid a bagillion dollars a year, and now everybody needs to take advantage – I’m sure that everybody saw the rumors on social media about Web 3, and how much people were making in that realm… I mean, that crashed overnight as well. So it’s a really, really crazy time, both for us in tech, but also I think outside of it as well.

Mark Nadal: Yes, so Firebase is completely proprietary and recently just got purchased by Google. They had a couple competitors, I don’t know if they were real-time or not, like Parse - they were bought by Facebook and just shut down. Open source to me is really important not only as I’m talking about the origins and the roots of the web, but also open source in a more specific setting.

For instance, a lot of databases out there are licensed under something called the GPL or AGPL license. Gun’s licensed under an MIT or Apache 2.0 or zlib license, which basically says “You can do whatever you want and you don’t have to pay us a dime.” Having technology available to the masses that they can use it without any concern or care - whether it’s for personal use cases or commercial use cases - is really important because if you look especially at the housing crisis in 2008, the entire world economy collapsed, but the one exception was the tech industry. The tech industry just kept on growing and growing. My argument for why is because it’s based on an open foundation. The web itself, in its protocol, whether it’s HTTP or other things out there, is open.

It might seem like nothing when you just have a couple university [unintelligible 00:27:40.10] coming together and saying, “Oh, here’s a proposal for how computers can talk to each other.” It’s so ethereal… It’s not like you’re moving gold around. There’s no tangible, physical resources to it. But because they came together and said, “Hey, here’s something, here’s an idea that has amazing potential, and we’re gonna give it away for free to anybody and everybody”, suddenly you get the rise of the Googles, the Microsofts, the Apples.

[28:10] You have to admit, Microsoft and Apple, or software or hardware companies - a lot of the money they’ve made is off of the services being internet-enabled and stuff like that.

You get the sudden rise of technology and growth and prosperity that keeps on growing, even during some of the world’s greatest recessions. That is why open source matters so much. The foundation of technology, if it’s based on, “Hey, I know you, you know me, or I don’t know you, but I’m gonna give you this for free, because I believe in this cause of decentralization and in open source” only helps things grow, and it makes the world and our lives better collectively. That’s why I’m passionate about open source.

Paloma Oliveira: I ask you because that also comes from my experience. I swear to you, I work five times more when I’m happy and when I trust the people I met… Like, oh, managers or projects I’d work for free, that I’d spend – I do not even need to sleep, because I’m so happy, and I feel nurtured because I believe in it.

If you put yourself into stress mode, like “Oh, we’re in a recession, we’re growing”, that stress just consumes you in a way that is so deeply that you cannot think. You just start to shrink. And that shrink is on your body; you see people are curving out… If you go from the perspective of the body, you see literally people shrinking out. You see people literally so stressed that they’re like in circles, in the same problem. An issue, and an issue, and an issue. You cannot see a light forward.

[52:06] Even in recession, even some economic theories will tell you “Open up. Do not close down, because it’s not beneficial.” So that’s not actually an argument.

Brian Ketelsen: There’s no money in that… It’s interesting, most of my older family members are in the restaurant business, and the restaurant industry has almost a Maslow’s hierarchy of types of food, and barbecue is the lowest demand food when there is a recession in place. People eat barbecue last when there’s a recession, so it’s not a recession-proof restaurant… So I would just never open a barbecue restaurant.

Besides, I think you’d lose a lot of the fun if you were cooking for a paycheck instead of just cooking for the whole neighborhood, like I do.

Shaquille Que: As I look forward in my career, I would say the most concerning right now is graduating in a recession. I hear a lot of stories from back in 2008, or even back in the 1980’s, when it was very difficult to find jobs and when a lot of people were losing their jobs… And I think it’s also already kind of happening right now, where a lot of companies are laying off and a lot of companies are also going under.

[28:02] I know that in the future the tech industry will still keep growing, and even right now, a lot of the big companies like Facebook and Amazon are still aggressively hiring… But it’s still scary to think that right now it’s gonna become harder to find jobs, and graduate in a [unintelligible 00:28:19.29]

Autumn Nash: So people were kind of freaking out, because this is like one of the first really, really big acquisitions since tech has kind of been in this tech recession… Like, there’s no more [unintelligible 00:55:16.17]

Adolfo Ochagavía: “The five things you need to know to avoid starving in the next recession”, you know? So just writing about normal stuff, from a real human perspective… And I’m filled with curiosity about computers, and I think I can pour that into my articles. And I also care about the people who are reading my articles. So I guess people, in some way, perceive that.

Autumn Nash: Okay, so “How the tech industry broke its promise to women.” The title is actually better than the link title. But I think it’s really interesting, because we dumped so much money into diversity. There are so many diversity programs. And it’s crazy, this is not something that is just warm and fuzzy. This is something that Harvard Business has done studies on over and over again. It says that you will get a better return on investment, you’ll make a better software, you’ll have better teams, and they’ll innovate faster. And this is all stuff that Forbes, Harvard Business, all these huge names in finance and business and how to build a business that are very well known have told us, “This is going to add more value to your business.” And just the amount of wealth that is going to happen in the wealth transfer to women over the next couple of years, economics are saying that it might be one of the biggest wealth transfers in history. And we’re “But we don’t need 51% of the population. We don’t need their inputs.”

And it’s crazy, because women actually – I think we dictate 71% to 80% of spending, and we influence that much spending, and I think 70-something percent, 72% or 76% of discretionary spending. So think about it, when these companies are selling products to people, you have one person who does a lot of the family spending. Or just in general is your customer. And you are not taking in their perspective to what kind of products are they looking for, what kind of testing should you be doing. And we not only sold women “Hey, come, be in tech, we need you”, and then we were “Just kidding, because money is an issue, and we’re in a tech recession, so we’re no longer interested in diversity.”

So what does that mean for tech? What does that mean for how we build software and our return on investment, if we have this thing that statistically shows we are not going to build as good of software, and have as good of a return on investment without having people from different backgrounds, especially women, who is going to be a majority of your audience, and the purchasing power, and then we’ve decided to just abandon that? Where do we go forward?

[00:06:11.08] Also, retention is horrible. It’s 56% by mid career, by 35. So not only is only 25% or 27% of the industry women, and that’s not even mostly a technical… I think black women are 1.6%, Latino women are 2%, and those are not widely, like, engineering, right?

Juan Pablo Buritica: Absolutely. I think what Julian says is spot on. We are also recipients of the butterfly effect that other people started. I personally did focus on JavaScript because of the community, but what predates this is when I got into technology around 2009, before that I was a line cook that ran a kitchen in Fort Lauderdale. I had moved to the States in 2007. The recession came along, I needed a job, I landed in the food industry, and I was there making my way up in the kitchen… But I also always liked building websites, and I had learned HTML and CSS because I set up the MySpace for my band, and I just – it was a hobby to learn… And I discovered the yayQuery Podcast, with Paul Irish and Rebecca Murphy and Adam Sontag and Alex Sexton, that predates a little bit this podcast.

[44:11] I would listen to them when I was closing the kitchen, and I would learn about if statements and just a bunch of random stuff, and despite the fact that I was learning (I think it was) Joomla in PHP, what I was actually building sites for people on, JavaScript kept calling me, and I heard about JS Conf in that podcast, the first time. Then they had a recap and were like “Oh, we went to JSConf etc.”, and I was learning about the internals of this community. I sort of felt that I knew each of them. Then the first JSConf I went to, I met Adam, I also met Rebecca, Paul was there… And now, more than someone who was in my ears, in a podcast, years ago, they’re folks who I consider friends.

Christina Warren: Yes. Well, so it’s funny, because my background, like going all the way back to when I first got into anything, when I built my first website when I was 12, I’ve always – for as long as I can remember, the two things I’ve loved the most in the world have been movies, pop culture, TV, and technology. Those have always been kind of my two biggest passions. And I was lucky enough when I was a journalist, and the story there is that I graduated from college in 2008, right in the middle of the Great Recession, and I’d always thought I was going to go to law school, and wanted to be either a technology attorney or an entertainment attorney; ideally, an entertainment attorney, because I figured “Well, I could leverage that and maybe be a producer, and work at a studio, or for a TV network, or something, and use my power to get projects I want to get made…” that was the big pie-in-the-sky goal. But in 2008, going to law school and taking out loans didn’t seem like maybe the best idea. It was a little bit – I think that I totally understand the hesitancy of kids who are graduating now, or who are going to be graduating next spring, you know, about the uncertainty in the world… And so I did the very logical thing, which - I’m being ironic here, because it’s not logical at all… And I was “Well, I’ll just go into the very stable career of journalism.”

Autumn Nash: I think the way that you’ve had such a perspective, Rich, with the fact that WebMD was kind of like around when there was that bubble… How do you feel now that you’ve gone through seeing these huge companies kind of come and go, and also seeing that first tech bubble? Do you have any advice for people, like for me and other people who this is your first time seeing this sort of like a tech bubble, and recession, or whatever you want to call it in tech? How do you look at it?

Jerome Hardaway: It’s okay. I was a glorified security guard/air cop. That’s what security forces is. You know, they do that weird thing, they’re like “Oh, here are the jobs that you can qualify for.” And it was like cop, security guard, air marshal, which ironically, none of those were hiring when I got in the military, because it was the Great Recession, so I was like “Oh… That’s awesome.”

Johnny Boursiquot: We’re all talking about recessions nowadays, but you’ve been in one for quite some time, right Matt? [laughter]

Adam Stacoviak: Yeah, I was talking to Jerod before this call, there was something I didn’t share in Changelog News, because I was skeptical of some of the details behind it, but it was this – the title says “Which one of these will be the biggest unicorn failure ever?” And it cites all these larger companies - Uber was one of them - with little to no profits, and they’re ten years old, twenty years old… And it’s a small swath of them… But there’s also tons of opportunity in the boring business sector. I mean, I’ve been turned on to this more half tempted to just to buy a bunch of laundromats kind of thing… But think about the innovation if you’re a software developer, right? I was telling Jerod this. This is sort of a silly, kind of crazy idea… There’s a lot of boom happening in the RV space, but there’s no Tesla in the RV space, aside from, say, Airstream. And they’ve been around for a very long time, and it’s mainly their quintessential external design, with good taste on the inside. Most RVs are kind of ugly, but almost none of them have a LAN network, almost none of them have high-quality software, almost none of them have a lot of thought put into like people who are younger and into more smart things, or automation things… It’s kind of like an untapped market.

If you can do electrical engineering, and mechanical engineering, and software engineering, and all these different things come to play in these boring sectors; or even recession-resistant places like laundromats, eldercare, healthcare, as you mentioned, storage facilities… There’s a lot of automation that can happen there. Like, if you owned an RV park that was automated fully by software - it takes land which appreciates, software, mostly automated, maybe one operator… Like, if you’re a software engineer, it might be time to follow, I think - what was her name, Jerod? …Ann’s advice from Wired? She was saying in the article you posted…

Dylan Bourque: Slightly related to that, speaking of Filippo and crypto, the regressions in 1.20, that he fixed in 1.21, and talked about at GopherCon this year - yeah, those were fun at work. We have a few systems that spend a lot of time verifying public-private keys, and things did not go well when that regressed like it did.

Nicola Murino: This is exactly the breakage I’m thinking about, because after the initial support we started to get reports because all the OpenSSH versions don’t work properly, because their bug, also some old version of gpg-agent, stopped working… And so we added a lot of new [unintelligible 00:50:37.19] Basically, our problem was that we have a senior interface that was unable to advertise the supported algorithm… So you can just assume that all algorithms are supported, but this is not applicable anymore. So we needed to introduce a new interface, a multi-algorithm to the senior, and advertise the supported algorithm so you know the supported algorithm and you can choose the one to use for signing. This is our way to fix the issue, because supporting the multi-algorithm signing allowed us to provide the ability to restrict and choose client-side, server-side, and also a certificate signing algorithm… Because one of the biggest issues with OpenSSH are the certificates that are a different standard from X.509 certificate; it’s something different, and this introduced a lot of issues with the old OpenSSH versions. Since a few days we merged the latest fix, so I hope we have a normal regression in this area… At least for a while.

Cory Doctorow: So what if we just had a lot of stochastically circulating, high-quality drills, that were location-aware, measure their usage, gather telemetry constantly for continuous improvement, were designed with the non-economic imperative of gracefully degrading back into the material stream when they reach the end of their duty cycle… All of this stuff that markets wouldn’t do, but that planning could, that would end up with a world in which you have significantly more material abundance. Like, anytime you need a thing, whether that’s – we’ve got like a couple of big folding tables and a whole bunch of extra plates for the one to two times a year that we have a ton of people over in the backyard when we set up a table and we have like a dinner party. And again, the square footage in this house is not free. The fact that I’m giving it over to this incredibly rare usage is not an asset, it is a liability.

So what if all of this stuff was just in circulation all the time? It could be in your neighbor’s house, it could be at the library, it could be somewhere else… And there’s just this kind of constant circulating abundance. I call it library socialism. And so this is all like IT-driven, telecoms-driven, communications technology-driven, smart coordination. It’s IoT without the terrible extractive business models and the planned obsolescence. It’s like what a people’s IoT would look. An IoT designed for sustainability, and not for extraction and enshitification. And that, I think, is quite an exciting way to think about the future.

[00:23:22.25] It’s easy to – look at all the stuff that we have that’s so bad in our technological realm. Everything has undergone such regression since 10-15 years ago. The quality of our goods, the reliability of them, the likelihood that they can be maintained, and so on. There’s just such incredible regression in product and service quality. And imagining that we didn’t have to have this entropic force of extractive capitalism bearing down on every technology that we use and rely on, so that like each new update isn’t potentially a way to smuggle a downgrade into a security update, like HP keeps doing, where it’s like “Oh yeah, this is a new important security patch for your printer. The security that it patches is that your printer can currently use third party ink. And once it’s patched, your printer won’t use third party ink. That’s about safeguarding our security; we don’t distinguish those. And by the way, we might also ship you a patch that keeps your printer from being infected with malware, and also call that a security patch. So if you don’t want to install security patches, you just go ahead, but don’t come crying to us when you print a rotten document and it rewrites the firmware on your printer, which then starts to like NMAP your LAN, run zero days on your computers and open a reverse shell to a command and control server that runs your whole network”, right? Which is an actual demo that [unintelligible 00:24:44.08] at one point. I saw him do that demo at CCC once.

Adam Stacoviak: Yeah. In regards to the drift, I think it’s – some have said in regards to remote work and sort of progression in certain directions, I wonder if there’s regression in the other directions, which haven’t really been talked about. So there’s been discussions around, “Oh, COVID and the global pandemic has just sped up the process of remote work. Like, it was going to happen. This, where we’re at right now with remote work - it was going to happen no matter what.” But this year, fast-forward 10 years - I wonder if some cases, or if some people have experienced a fast-forward into burnout; a regression, essentially, of balance, because of the state of the world. Because I know this year has been uniquely hard for many reasons, but I feel like there’s been some acceleration in the regression state, for some things for me even. And it’s like “What did I personally do to allow that?” You used the word “tether”… You know, “Where did I not tether properly to disable those things from happening?”

Eric Normand: The reason I think people ask this question a lot is they – like we were saying before, they hear of functional programming as a list of features, and they’re like “Well, I have functions, and I have data, and I have recursion in my language… What does some functional language give me that I don’t have?” and they want some magic feature that you need when you switch over. It’s almost like a challenging question, it’s like “I know functional programming, because I know how to make my data immutable, I know how to work with that; I know how to write a function, I do callbacks all the time.”

So the question is more like “Why can’t I just do functional programming in my language? Why should I learn Clojure or Haskell?” My answer to that is yeah, you can do functional programming in any language, just like you can do object-oriented in any language, including C. You write your own little object system that passes messages and encapsulates state, that kind of thing, but are you really gonna learn object-oriented programming by learning C, by staying in C? Probably not. When someone actually does develop an object system in C, it’s probably because they’ve been versed in Java or C# or some other object-oriented language and they’re kind of back-porting that into C, because that’s what language they have to use.

So I have the same answer - if you want to program functionally in any language, sure, but it’s not gonna give you any help for learning it, and to just think of it in terms of a few features, such as MapReduce, filter and those kinds of functional tools, you’re not gonna really get that holistic perspective, that holistic approach to a problem; you’re just gonna sprinkle it where it seems convenient, and it’s gonna be too easy to just back out to the stuff you’re already familiar with.

[33:52] It’s just like learning a spoken language - you wanna immerse yourself in the language, so you set up conversations with a buddy, so you can practice speaking. “The first rule is we cannot use English, right? We can only speak French for this hour”, or whatever language you wanna learn. If you don’t have that rule, you’ll notice by 30 minutes in you’re mostly speaking English, with a couple of words in French, and you need to just like jump in, jump in the deep end.

Jerod Santo: The curse of knowing how

Raf beautifully describes the plight of the enlightened who don’t just use computers, but have the ability to program them:

Before I could program, broken software was frustrating but ignorable. For years I’ve simply “used” a computer, as a consumer. I was what companies were concerned with tricking into buying their products, or subscribing to their services. Not the technical geek that they prefer to avoid with their software releases, or banning from their games based on an OS.

Now it has become provocative. I can see the patterns that I wish I couldn’t, find oversights that I can attribute to a certain understanding (or the lack thereof) of a certain concept and I can hear what has been echoing in the head of the computer illiterate person who conjured the program I have to debug.

This is problematic for a bunch of reasons. One of which, as Raf states, is that pesky thing called entropy:

Software doesn’t stay solved. Every solution you write starts to rot the moment it exists. Not now, not later, but eventually. Libraries deprecate. APIs change. Performance regressions creep in. Your once-perfect tool breaks silently because libfoo.so is now libfoo.so.2.

Sadly, just because we can fix computers, we can’t fix everything. Even with computers. So what’s a dev to do?

After the excitement. After the obsession. After the burnout. I’m trying to let things stay a little broken. Because I’ve realized I don’t want to fix everything. I just want to feel OK in a world that often isn’t. I can fix something, but not everything.

You learn how to program. You learn how to fix things. But the hardest thing you’ll ever learn is when to leave them broken.

And maybe that’s the most human skill of all.

Daniel Whitenack: One of the things that you mentioned is this evaluation phase of what goes on as helping prevent regressions, because in sort of testing behaviorly the output of the models you might make one change on a small set of examples, that looks like it’s improving things, but has sort of different behavior across a wide range of examples… I’m wondering also, I could imagine two scenarios… You know, models are being released all the time, whether it’s upgrading from this version of a GPT model to the next version, or this Mistral fine-tune to this one over here… I’m thinking even in the past few days we’ve been using the Neural Chat model from Intel a good bit, and there’s a version of that that Neural Magic released, that’s a sparsified version of that, where they pruned out some of the weights and the layers to make it more efficient, and to run on better – or not better hardware, but more commodity hardware, that’s more widely available… And so one of the questions that we were discussing is “Well, we could flip the version of this model to the sparse one, but we have to decide on how to evaluate that over the use cases that we care about.” Because you could look at the output for like a few tests prompts, and it might look similar, or good, or even better, but on a wider scale might be quite different in ways that you don’t expect. So I could see the evaluation also being used for that, but I could also see where if you’re upgrading to a new model, it could just throw everything up in the air in terms of like “Oh, this is an entirely different prompt format”, right? Or “This is a whole new behavior from this new model, that is distinct from an old model.” So how are you seeing people navigate that landscape of model upgrades?

Natalie Pistunovich: Just in your reception hours. “Here’s a rubber duck. Speak to it.”

Dave Lacey: Yeah, exactly that. There’s a lot of investment – well, one is the development… A lot of software developers [unintelligible 00:30:41.05] a lot of investment in the regression system, and test infrastructure… And it all has to be done, because you have to have a really robust, comprehensive software product. There’s no way around that for it to be usable. And to kind of blow my own trumpet a bit here - I think Graphcore, amongst the graph of AI startups, is really well advanced in that space. All the elements you have [unintelligible 00:31:03.07] the documentation, and how you do that, and so on.

So I think we have to try and keep on top of that, keep working internally at Graphcore to keep those [unintelligible 00:31:14.26]

The other thing I’ll say that kind of really helps with this is being very open… So making sure that people can – we document, and make sure that people can access well… This is something that you do get on other platforms, but I think we’re leading in this, and being able – to know what Poplar, the low-level graph compiler, explaining that, making sure that people can add custom operators in that… Those libraries I talked about - they’re open source on GitHub… That kind of thing. Our TensorFlow and PyTorch backends are open source and everything, as well.

So by having a more open infrastructure, it makes it easier for the community at large to help you adapt to new things as well. And as we’re getting more popular, more people using us, we get more community involvement like that as well, and I think that’s an important part of it as well.

Jerod Santo: Sorry, I went to a football metaphor, and then I immediately lost it. So we should be upfront with that. This is not an easy discipline. It’s gonna get really hard, you’re gonna hit this problem, that problem, and the other thing, and you’re gonna have to be able to power through that. There are not just complex issues, but even just topics that are really hard to grok - recursion, pointers, computer sciency things.

So that should not be like “Hey, everybody should be a developer, because it’s easy, and it’s fun, and it’s all unicorns and rainbows”, because that’s just not the case. There’s a lot of hard work, and all the things that you’re saying, Kris.

That being said, the quick wins are very empowering, and easy on-ramps aren’t gonna stop the people who are gonna make it through. The ones who did really good at the recursion class in the class - they’re also gonna get through those easier on-ramps just fine… But what those on-ramps can provide for somebody is sort of like an extrinsic motivation that they may not have had otherwise, especially for young people, that hey, there is satisfaction in this career, there is empowerment in this career, there is things that you can do which will give you joy, or help others… And you can do those things if you power through, and we can get you started as easily as possible to get to the hard parts. I think that those might provide some people with motivation that they may never have gotten to if you just hit them over the head with recursion.

So I’m kind of of both minds. I see what you’re saying. We shouldn’t bait-and-switch, but I don’t think that means we have to make it hard on purpose, because it’s already gonna be hard. Or not easier… I don’t know. I can’t remember how you worded it.

Matt Brems: Sure. I think that probably the other biggest misconception, at least what I see from students that have come in on day one and who know the broad idea of data science, but the inner workings… That’s why they’re coming to General Assembly, to learn a lot of that. And this is certainly not a – everybody comes in with this misconception, but something that people are so much surprised to learn is how much data science you can do without getting into that artificial intelligence or that neural network kind of level of doing data science.

[15:55] When I use the term “artificial intelligence”, what I mean is we’re trying to get computers to mimic or to simulate human intelligence. I think that people are surprised when we teach models like a linear regression model. For listeners who may not have hear of a linear regression model, you may have heard of the term “line of best fit” in the past, where you’ve just got a scatter plot of data points and we need to put a line through that data. Or thinking about a slightly more complicated model, but a related one - a logistic regression model, where you’re using a curve of best fit to try and prevent a one-zero outcome. For example, for my political experience, will this person vote or won’t they?

I think that the misconception for a lot of people - or a misconception - is that people often think that neural networks will be the solution to all of their problems, when in reality linear regression, logistic regression, much more basic techniques (for a lack of a better word) are really helpful and are often the solution to the problem that you’re trying to solve.

Daniel Whitenack: You mentioned that each of these nodes or neurons is fully connected in this network, and each one has its own inputs and outputs. Now, if we dig into one of these neurons to think about what’s inside of that bubble - and again, you can think about that visually, like a bubble or a node, or if you’re a programmer you might think about it as one of these sub-functions under a big function… But it has its own inputs and outputs, and if we think about it maybe as just having a couple inputs - let’s say X1 and X2 - what happens inside of that circle, or inside of that neuron? Well, there’s some kind of simple things that happen often.

One way we could think about processing these inputs in the neuron is to just add them up. In a linear regression sort of way, we could multiply each of my inputs - X1 and X2 - by a couple coefficients. Let’s say W1 and W2. Those are often called weights. So I just add up the two things after I multiply them by coefficients, and then I might add in an intercept or a constant, so just X1 and X2 plus something; that’s often called a bias.

In this case, I would have three parameters that parameterize the way I’m combining these inputs. So when each of my X1 and X2 come in, I combine them together in this way. And that’s all good and fine, except most relationships in our world aren’t linear, so it might be good to introduce some non-linearity into this combination. That’s where a thing called an activation function comes in… Which is just a non-linear function that’s kind of applied to this combination of inputs to give it some non-linearity.

Common functions that are used are sigmoid, or ReLu, or hyperbolic tangent etc. that are applied to this combination of inputs. So when my inputs come into that node or that circle, they’re just added up in a special way, and then output out the other end. All of my neurons in my network kind of do similar sorts of simple operations that are parameterized in a similar way. So each neuron has a certain number of inputs. They’re combined together using some parameters, and then output as a number. That’s what each neuron does.

David Heinemeier Hansson: So the first release candidate is out. We pushed that out on Friday, I believe. I think, if it’s not going to be this week, then hopefully it’s going to be next week. Unless we find something terminal… Well, we’re never going to find anything terminal. Unless we find something serious that requires a long time to fix and I can’t see it, because all the major – well, not all. Many of the major Rails apps, they’ve been running Rails 8 in production for a long time. Shopify runs Rails 8 in production, GitHub runs Rails 8 in production, we run it for Basecamp and Hey… This is the dividend of modern Rails development, is that the new versions, they’re generally exceptionally solid, because they’re being run early at such an incredible scale that if there is a problem, Shopify is going to find it. GitHub is going to find it. We’re going to find it. Either in ergonomics, because there’s a lot of developers who have eyes on it, or in terms of performance regressions, or outright bugs, or something else like that. So we’re just being extra safe here. Extra careful, which I think is good.

The quality has risen immensely. Those early versions, version 1, version 2… I mean, maybe you wanted to wait for the 0.1 version for that to be a little safe? That’s not the case with modern Rails, and it hasn’t been for a long time.