Zeno Rocha: I already saw there’s npm packages for the purchase parity power (PPP) concept.

Jerod Santo: Nice. And so I see now perhaps your bottle caps idea moving over to this, because now we have doubloons. This is basically the equivalent of your – this is your purchasing power here.

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?

Amal Hussein: So you have to understand, the difference with iPhone users is that they represent the wealthiest people in the world. It’s very different – like, I don’t know, is it 16% of the global market share, 18% of the global market share is iPhones? It’s under 20%. But that 20% represents the majority of the world’s kind of purchasing power. Does that make sense?

Asim Aslam: I’m with you there. I think it’s one of those things where you can kind of talk about one side of the equation and not the other… It would be great if that was an option, that was available. At the same time, who knows what would happen if it did exist, and all the different things that would occur?

When you do have a brand like GitHub you can’t just launch something as a test anymore. It has to really be a real thing. I remember before they started doing sponsorships they had some people reaching out to some of us in the open source community, those of us who had prominent projects, to have these kinds of conversations, like “Hey, what are your struggles? What do you need? How could we help? What do you think should exist?” And sponsorship was one of those things that I mentioned, like “Hey, I’m having to use Patreon, I’m having to use separate things from a corporate sponsorship. If you just made it really simple for people to be able to sponsor, that’d be a great way to get donations, and that would work.”

To their credit, they eventually ended up rolling that out. That’s working for some people, and that’s phenomenal. They rolled out Discussions in the same way, and various kind of new features… And I wouldn’t be surprised if they were working on it. I just have this feeling – you know, I’m always early to things, and then I feel like I miss out… So I feel like I was early to this solo open source thing, and I feel like in 4-5 years you’re gonna be able to charge for your software on GitHub, and people are gonna make real money. And it’s not gonna be like “Hey, please sponsor me.” It’s like “I’m producing a piece of value-add software, and you want my support, or you want these additional features”, and GitHub has enabled a way to make that seamless, to provide people purchasing power in such a simple way. And that will happen. Then there will be someone who is making a million dollars a year from being a solo open source creator. That’s something I truly believe will happen in the next decade.

Gergely Orosz: Well, because it’s a topic that I think is pretty fluffy, I wasn’t comfortable with launching something that is just halfway there. And also, one of the things – I really was debating whether I should charge for the book or not… Because I’m a pretty prolific blogger, I write a lot of blog posts, and people really get a lot of value out of it… I wanted to dry-run this theory of – my goal with this book was to get engaged readers.

I have some blog posts that are read by a hundred thousand people, but I never know if they’re engaged or not. So my theory was “Well, it’s a lot of research and a lot of effort has gone into this book. I’m trying to price it at a decent range.” It’s $20 in the U.S. and across the world I have purchasing power parity; for example, in India it will be 60% off. In a good part of the world, similarly… So based on your IP, it’s a lot cheaper.

What I’d like to know is how many people are engaged… Because what I find myself is if I buy a book, which is usually decently priced, like $20, $30, $40, I actually end up reading it, because I spent the money. And I wanted to test the theory of like “Can I get engaged readers?” and the engaged readers are either people who buy it or people who need it. And the people who need it most, they can get it for free, and I assume that they’re reading it.

So far, the theory is going pretty okay. I have a bit over 2,000 engaged readers, I should say. So either people who bought it, or are using it… And I’m getting a lot better-quality feedback as well. Some people are giving me specific feedback on what they’ve seen. And so far I have zero refunds, which I think is – usually, from what I understand, eBooks usually have a full-refund policy; if you don’t like it, you get your money back, no questions asked. No one’s asked for that yet. Usually, there’s like a half percent in the eBooks business for similar authors, from what I asked around.

Quincy Larson: I can say from our experience our budget was literally zero. We’ve spent nothing on this conference. And the reason was we were able to use royalty-free music from Base Rebels, which is a really cool livestream on YouTube, and then we have it some other places… But it’s all music that is either public domain, or licensed to where you can play it in the background.

And I think some other – Suz would probably know a lot more about this than I do, but MonsterCat and some other radio stations you can play in the background and they won’t get muted or they won’t block your video on YouTube or Twitch.

So we were able to get music from that… And then we have this gentleman, Matthew Potter, who I met him ahead at a FreeCodeCamp conference in Toronto… He has volunteered a tremendous amount over the years, and he put together all the graphics… And he’s like a designer who can program, who also knows a whole bunch of video production stuff. So he was in the control room, making sure all the feeds were patched together… And we just used Skype, because they have a pretty robust API, and he was able to pull the feeds from Skype and put them into his system, and have this fancy UI, I guess…

There were the little windows with the different people chatting, and then he could do – like, if one person was talking for a long time, like I am right now, you could just pull up just an image of that person talking… And then if two people were talking back and forth, you could put two side by side and hide everybody else… And then you could zoom out and have all five people - the two MCs and the three panelists on the screen at the same time. He just did all that stuff…

So unless you have a friend like Matthew Potter, it may cost you some money for those things. But for us, it cost us zero dollars. And we didn’t even have to really moderate the chat, which is normally something he had to do, because you get a lot of low-quality comments, and unfortunately the reality is if you have a woman on the stream, people will always be like “Oh, she’s pretty”, and things like that, and you don’t want that kind of stuff… So we didn’t have to do that, because we had members-only chat, and people had to pay like $5 or 200 Rupees or different currencies (this is scaled to purchasing power parity), so that reduced the overhead in terms of the amount of attention we had to pay to chat, and the number of people we had to have moderating.

Adam Stacoviak: That’s right. Steal everything. You heard it here first. And I suppose that those who are trying to sponsor/donate, the same call to action: ladybird.org. Find a button to donate, find a button to sponsor… You’re keeping all these sponsors unrestricted… They don’t get board seats; you can’t buy a board seat. There’s no direction that you can purchase. You can simply support and be promoted as someone who cares about supporting Ladybird. That’s the whole purpose of a sponsorship; it’s not buying a board seat or getting power, and having the clicker.

Bryan Cantrill: I know, right? “We’re going to take the technical excellence of Sun, and we’re going to combine it with the business savvy of Oracle.” I mean, it was almost immediately after the acquisition that I realized just how stupid that was, and that Oracle doesn’t actually have business savvy in any true sense. Oracle understands coercion. That’s what Oracle understands. And the way that Oracle operates is they get the leverage to be able to coerce their customers, and then they lack the mirror neurons to stop themselves from just excruciating coercion.

I felt embarrassed to work for Oracle, which is not something – I felt ashamed. Not something that I would have anticipated. But one of the things that I really learned very quickly after the acquisition is despite Sun’s faults - and Sun had plenty of faults - Sun’s customers, our customers wanted Sun to succeed. And what I realized was like, despite our faults, our customers trusted us, because we were always transparent with them, like warts and all. Sun did try to pretend that it was something – I mean, again, every company has got faults, but for its size and history, and staying power, Sun had really surprisingly few… I mean, no one went to jail at Sun, there were not these major ethical transgressions… And Sun’s customers, by and large, wanted Sun to succeed, because they trusted the company. And that was not true with Oracle. When you get to Oracle – Oracle’s customers don’t trust Oracle. And Oracle’s customers are – I mean, when I was there in 2010, the state of California refused to purchase any more Oracle. The state I lived in, the government of the state I lived in refused to purchase the product, because they had outstanding lawsuits… I mean, that’s not what I want to do with my career.

[00:44:40.03] What I learned in that whole process is that the meaning of an endeavor is really important to me. There’s got to be a bigger why. Why are we doing this? And I have always believed that in computing there’s this line – I’m not a huge Halt and Catch Fire fan, but there is a line that I do love, which is… I’m a much bigger Silicon Valley fan than Halt and Catch Fire.

Jerod Santo: Herman Õunapuu calls Zimaboard “the closest thing to my dream home server setup”. What’s a Zimaboard? “A small single-board computer that is relatively affordable and comes with an interesting selection of ports, which includes an exposed PCI Express port.”

Herman wrote up a full review. It starts with: “What sets the Zimaboard apart from other single board computers is how polished the product feels. Unlike a board like the Raspberry Pi, this one comes with a case and a cooling setup already attached to it. I suspect that a similarily configured Raspberry Pi 4/5 with all the accessories added on top would result in a price that’s quite similar to the cost of a top-of-the-line Zimaboard.”

He goes into its storage, power consumption, performance & caveats, which is always nice to see in an otherwise glowing review. His conclusion: “Overall, I’m very happy with this purchase. The performance is just enough for my services to work reasonably fast, the board uses very little power and it’s silent, and it looks good on my wall.”

If Zimaboard is news to you and you have a dream home server setup, give it a look.

Daniel Whitenack: Yeah… I mean, I think if you draw a parallel with what’s happened with other models that have caught on in similar ways… Like, if you imagine back to BERT, and these large language models, part of the magic of those was that the weights were open source, you could pull down a pre-trained version and then fine-tune it for a particular task, right? So I have no doubt that – and I think people are looking into this, and… There’s explicit notes on the Stable Diffusion page about limitations, and bias, and all that. So you can read that there.

But certainly, there’s bias in the dataset on which it was trained. But I think the power comes is - if you’re able to open source the model, in some sort of way, with tooling that will allow for the fine-tuning of it, I’m sure that people will sort of fine-tune or create different versions based off of the parent using maybe its imagery for particular styles of books, or publications, or imagery, or inpainting for creative arts, or for video processing, or for all of these different things. I think people will create their own versions of these, and probably some of them will be – those fine-tuned, kind of purpose-built models will be commercially available for purchase, as we’ve seen with certain language models in the marketplace… And some will be open source for people’s usage. Just like we’ve seen kind of a general-purpose BERT, and then we’ve got like a science document BERT, and we’ve got a legal document BERT, and these sorts of things. And those are open. But also, there’s companies that are making money because they’re processing legal documents with BERT, and they have their own proprietary version; or maybe they’re using the open source version and just have good tooling around it.

Jerod Santo: So it seems like - okay, so the Raspberry Pi 4, if you bought like a maxed-out one, would it be the kind of purchase where you could run three, four, maybe five different automation things on it? Does it have that kind of power?

Brian Douglas: I’ve mentioned earlier I’m cheap, but I did splurge last year on a new camera. It’s a combo - it’s the camera and the lends, and I think the lens is really what makes the shot that I have right now. If anybody’s seen me stream, or been on a Zoom call with me, you understand the [unintelligible 00:51:45.02] that you’re getting right now in the pizza… But it’s a Canon EOS M200. The M200 is basically the same thing as the Sony A5 or A6 series. I think the Sony ones get recommended much more, and the reason for that I think is because you could plug in a USB and it would power from your laptop or computer.

[52:07] The Canon - you do have to have a dummy battery. I’m not a Canon fan, but my wife already had a Canon camera, so it just made sense to get the Canon instead of the Sony, at least memory cards transfer, and stuff like that.

But the thing that really makes it is a Sigma 16mm lens. It’s a lens that – you can’t see it in the podcast, but it’s about maybe 4-5 inches long, and it actually gives you that sort of… Far enough from enough from your face, but it’s big enough to actually give you that [unintelligible 00:52:32.01] YouTube-esque feel as you’re doing Zoom calls.

The camera was about $499, or around that number, and the lens was about $300-something… And like I said, I splurged on it; I did not want to buy this, but it was on my wishlist for the longest time, and then I saved up enough money and got a bonus, and I was like “You know what - I think I’m gonna go ahead and purchase this thing.” And I actually have not regretted it since.

Daniel Whitenack: Gotcha. So you have the commonalities of the language and how words are used in the language… It sounds like your goal in developing the unsupervised method is then to take those steps lower where what you’re really trying to develop are the probabilities or the power laws associated with the words associated with reservations, or the words associated with purchase, or with making an appointment, or whatever the thing is.

So you have these sets of calls that have a certain more distinct statistical relationship amongst themselves that allow you to separate them out. That’s the goal that I’m picking up on… Did I say that right?

Adam Stacoviak: So in terms of theming, I think for me, I like Dracula. I’ve supported - although I don’t think he needs much support with it now - Zeno, with his new company, Resend. Dracula Pro. So I’ve been a Dracula Pro, buyer, subscriber, purchaser, whatever you want to call me… And I will almost unanimously go into my settings and do that. And so for me in particular, with my – I’m a Terminal app user, and I’ve got Dracula Pro setting. I think I’m using Van Helsing, or something like that. So I’ve like literally pulled the theme into OhMyZsh, configured it in, even fine-tuned it a tiny little bit, just because there was one thing that was bugging me about it. But that’s the kind of user I am. I want to be able to select some themes… But again, what you’re trying to do is make that hard path easy, make everybody a power user, and so I’m sure you’ve got – I see these here, you’ve got like Solar Light, and Dracula…

Scott Hanselman: Oh, man… I was trying to explain to my 17-year-old the difference between a phone purchase and a laptop purchase.

You know what I mean? Where it’s like, “Hey, we’re going to go to Hawaii. Okay, I’m going to go ahead and book the flight.” Like, what are you doing touching your phone? That is a laptop moment.

Preston Thorpe: But yeah, when I came here, they have – believe it or not, actually, the state of Maine, and from what I hear, most facilities… It’s probably something that viewers will be interested to learn. They have quite a bit of internally written software by residents. That’s actually a fairly common thing. And I definitely was not aware of this prior, but the Main State Prison had a resident there that wrote pretty much all the software that it runs on to this day. So all the account sheets, the inventory for the woodshop, and inventory for the kitchen, and a system where you scan a barcode in the back of your ID, and use that to access your account to purchase things from the commissary… So there was quite a bit of this handrolled infrastructure at the other facility. And the administration here was aware that that existed. And when I came here, they had been made aware that I was someone that could replicate this. And definitely someone that could maintain some of this stuff here.

[01:10:04.06] So the system, the point of sale system that they use for the [unintelligible 01:10:09.08] is written in – it’s vb.net. Let’s say it’s VBA. Yeah, it’s definitely VBA, because it runs on Access, and Windows 8. So to this day I still maintain a point of sale written in VBA, on Windows 8, which is the absolute bane of my existence.

But yeah, so the staff here asked me to build a real-time – something to keep track of what everyone’s position in the facility is. So when you come back to the unit, you scan your ID, and it will check you in to the units. Because currently you have to sign out, and there’s people whose jobs it is pretty much to maintain these clipboards with everyone’s current location, so that if something happens, they have to count everyone, they have a record of where everyone is. So they wanted that built. So yeah, I built that RFID-based system; I made that pretty much production-ready. And that – they pretty much abandoned wanting to actually get it set up, because I think they were just… The guy who wrote all the previous software at the other facility ended up getting out, and now they’re in a position where they don’t have anyone to maintain it, and I think they’re just trying to avoid that happening here. But I’m actually pretty grateful that they had asked me to do that, because writing a full production front, full stack application definitely I think is being on the list of projects that I completed by myself. I definitely credit quite a bit of my ability today to that.

Jerod Santo: That’s the news for now, but go and subscribe to the Changelog Newsletter for the full scoop of links worth clicking on. Such as:

• Self-host your own open source AI research agent
• A web-based, free alternative to Screen Studio

Get in on the newsletter at changelog.com/news

Oh, and I did want you to know you can
****now buy all of our Changelog Beats albums as digital downloads direct from our Merch Shop! And (for a limited time) use code BOGOBEATS to get a FREE album with each purchase!

Have a great week! Like, subscribe, and leave us a 5-star review if you dig the show, and I’ll talk to you again real soon.

Jerod Santo: Did you actually put a purchase down?

Ilya Grigorik: I think it’s context-dependent. I think there’s definitely a class of commerce in certain types of transactions where I know exactly what I want, there’s very low risk, and it’s kind of a predefined flow where I just say “Look, I need another carton of milk. You know exactly what I’m looking for, you know where to shop, and please go finish it. I just want it on my front door.” And then there’s other types of experiences where maybe this is your first time engaging with a merchant, maybe merchant has a set of rules where they actually require the agent to decelerate, because “Hey, for compliance reasons, I may need to verify your age” or “I need you to read this disclaimer on this product before you purchase it.” You can’t just have the agent blindly ignore that context, or click approve, and then proceed with the transaction.

[00:59:56.20] So I think we’ll need to define some protocol or shared mechanism to signal to agents that “Hey, in this particular case, I need you to pause and ask for a human to either confirm or take over control and complete the transaction.”

John Long: …too much. But for picking it for a project, to me, I wouldn’t want to use a site like that, because I don’t want to just – like, I have to negotiate with my boss to purchase a new icon set… That’s the last thing I want to do. But if I can say, “Hey, it’s 30 bucks.” It’s like “Great. Do it.” That’s 600 icons you don’t have to design.

Jerod Santo: An app to lock your apps until you touch grass

Rhys Kentish wanted to change the habit of reaching for his phone in the morning and doomscrolling away an hour, so he built an app to help him do just that. It’s built in SwiftUI and uses the Screen Time APIs provided by Apple plus Google Vision to recognize grass or not. Cool idea, but tough to manage for anybody living in deserts or tundras. Should he add an in-app purchase to change touch “grass” to touch “snow”, “dirt”, or “sand”?

Techno Tim: Oh, yeah. It absolutely could. Because you could use that GPU for Ollama or for anything else while you’re not using it, or even while you’re using it. Because it’s just going to use the CUDA cores if you’re using NVIDIA. And so it’s going to use CUDA cores and VRAM. But if you’re typing a document, you don’t need that. You don’t need any of that, you know?

So yeah, you could do both. You could have it run both. And then you could keep the AI local. And I think there’s desktop applications that you could just install it and do it all local, local; not even on a server in your home, but on the machine you’re using. So yeah, that would totally work. It would totally work.

3090 - yeah. I have a 3090. I got pretty lucky, because right before the pandemic, or right as it started, that launched, and everybody wanted the 3070 and 3080, I think. And I wanted the 3080, but it just so happened Best Buy had one of those in stock, the 3090. And I thought, “Oh my gosh, I’m going to spend like a thousand dollars on a GPU?” It turned out to be the best purchase ever because you couldn’t find them after that. And I used it all through the pandemic, for all of my videos and everything. It was great. But I still think they’re solid.

I like the founder’s edition. If you’re going to buy something, I feel like the founder’s edition directly from Nvidia, just the design and everything, is fantastic. I understand why people don’t choose that, but that’s just me; my personal preference.

Adam Stacoviak: This Digit though… So if you were to purchase it, what would make me think it’s valuable is if this thing can be on my network and give me local LLM to my LAN, maybe I can assign a global domain name to it, and actually access it from external, I can tap into it from the API… Maybe I can attach a client to it that fetches via the API, and it’s all local to me. Probably I don’t have to do what you said before, which is other people’s robots, OPR…

I’d like to own my own robot. And if Digit is the way to go, then it’s… Three grand to do that. Maybe the Gen 2. Always buy the Gen 2.

Phillip Carter: Yeah. Unstick you a little bit, and fundamentally encourage curiosity. Because if we get you to dig deeper, get more curious, ask more questions, get more assistance that way, that makes you, frankly, a better user of Honeycomb. It’s really good for our business. There’s a direct line that you can draw between how many users in an organization that has purchased Honeycomb are actively engaging with the product, and how likely that organization is to want to purchase more Honeycomb, re-up their contract, maybe buy even a whole lot more… And it’s like “Alright, cool. We have –” We managed to make it such that our business aligns on encouraging people to be more creative. And I’m like “Great. This is perfect.”

Matt Colyer: [00:23:46.07] So we’ve got several products that are AI-enabled or magic-enabled at Shopify. So Sidekick is kind of the main one, which we’ll talk more about… But to give you a general idea, it’s a tool that helps merchants find their way around Shopify, but also answer questions about their business. So you can think of it as like the co-founder they wish they had, that’s available 24/7 and isn’t judgmental. So that’s kind of like the Sidekick idea. And then we’ve got a variety of other ones as well. So we have a lot of imagery… So it turns out shopping - people want to see what the thing is before they purchase it; not terribly surprising. And one of the things that merchants often want to be able to do before they’ve scaled up to a whole team that has a studio, and a photographer, and the rest of it, is they want to enhance the pictures that they do have. So at the scale that they’re at - this is where technology, again, bringing back the best from the frontier, and making it accessible to all of our merchants is exciting.

So there is technology that’s out there now, that you can essentially describe what you want to do to an image… You’re like “Hey, my background’s a little bit messy. Can you replace it with a studio background instead?” Because we all know that the nice, white studio background that looks like the object’s floating in space. You can do that in real life, it’s just really hard and expensive, and you have to know what you’re doing. And there are very few people who know how to do that well. And so it turns out – we’ve created models that can do it fairly well as well. And so bringing that technology back - that’s one of the products we do offer integrated into Shopify today, is background generation.

So merchants can import an image that they already have, replace the background with something that’s more on brand… Like, say they want to set their coffee to the background of a jungle, right? They can place it on a table in front of a jungle, if that’s what they would prefer. Or they could do it into the void of white space. So lots of exciting opportunities with that.

Another area that we’ve been investing in is that we have a product called Inbox, which allows our merchants to talk with the buyers that they have on their site, and if a buyer has a question, like “Hey, what’s your return policy?” or “Where’s my order at?”, they can interact with the merchants through Inbox. And so one of the things that we’re offering today is that we look at all the merchants’ policies, and all the other things that they’ve given to us, and then can help formulate answers for those common questions. It’s like “Well, what’s your return policy?” It’s like “Well, we’re pretty sure that this is the answer.” And then we can suggest that to a merchant, who then says “Yup, that’s right.” Or if that’s not right, they can adjust it to be correct, and then send it. Merchants love that, because it saves them time for answering a lot of those repetitive questions. And then those ones that are a little bit harder, they can write them themselves just as they would before.

And then the last one that I’m thinking about is – or again, going back to that product and merchandising kind of task, is that oftentimes merchants are uploading a lot of these at the same time. So they don’t always capture all the metadata in the first go… And so this is, again, where we’ve created models that actually can help with that.

So if you upload an image of a white flower dress, we have a model that can actually understand what that picture is, and suggest that “Hey, maybe this is a white flower dress”, and this should be categorized under dresses, and under Cotton. And maybe it can suggest “Oh, it turns out –” If you upload multiple different colors, it’s like “Well, maybe you want to create product variants.” So that’s some of the other technology that we’re kind of working on today, is using the data that we have from merchants, enabling them to more expressively describe their products through our sites.

Avindra Fernando: [01:00:25.09] Getting started with it. Learned a ton there. Constant innovation, constantly grinding… Great, great, great time there. What I was thinking to myself at that time was “Okay, now I’ve got the product startup perspective. How does services or consulting work? Let me go experiment that.” So I joined a services startup, which their motto was consulting. A couple of guys, amazing, amazing dudes. Got to work with them, see how they negotiate contracts, bring in different contracts…

One of the contracts was so interesting to me. I was working on an app for someone; that was his hobby. He wanted this idea of a virtual bar. So he was mapping out all the bars in the cities that they go to, and would give the ability for someone to purchase a seat in the virtual world, which was a fascinating idea. I was like “People pay for this stuff?” It’s like, yeah, this is cool stuff. So I got really motivated by that.

And then I eventually decided, “Okay, I’m just going to start this journey on my own and see how things go.” So that’s fast forward to 2021.