"The Economics of Programming Languages" by Evan Czaplicki (Strange Loop 2023)
Вставка
- Опубліковано 5 жов 2024
- In the mythology of open source, programming languages are created by people who seemingly have no direct economic function. They are just really good at compilers (somehow) and have a house to live in (somehow) and have a lifetime to devote to creating a useful programming language (somehow!)
We will examine specific organizations that create programming languages. Where do the salaries for compiler engineers come from? How does Go end up with 5 engineers and Dart end up with 30? Who signs off on these expenses and why? Does this put any boundaries on language design or development practices? And how do the economics work for people outside of the major tech corporations?
My goal is to give the talk I needed to hear 10 years ago when I was just starting on Elm. By clearly delineating the many variations of corporate funding and independent funding, I hope users will come away with a better foundation for evaluating and comparing programming languages.
Evan Czaplicki
Creator of Elm
@evancz
Evan is the creator and developer of the Elm programming language. He lives in Denmark, working alongside his wife at elm.studio to keep Elm independent and interesting.
----
Recorded Sept 22, 2023 at Strange Loop 2023 in St. Louis, MO.
thestrangeloop...
This talk: "Makes me laugh, then makes me think... then makes me cry". Because we as a civilization haven't found a way to reward people for being altruistic.... and sharing with others.
The main point of being altruistic is to not getting something in return. In fact you get something which is a good feeling which would lead me to the conclusion that even altruism is selfish.
@@cypherecon5989 no, you can't mess specific and rare **altruistic selfishness** with (very spread *egoistic*) selfishness
Call me crazy but I would have thought the main point of being altruistic was improving the situations of other people.
Is it the goal of capitalism? 🤔
Elm is a beautiful language and a joy to use. It's sad that Elm has not flourished and Evan has not found a way to make a happy career out of it. His points about big tech being rentiers and the threat of being "Jeff'ed" remind me of what Cory Doctorow has been writing about the need to break up big tech monopolies and force interoperability.
I recommend reading the book "Start Small, Stay Small" - it's been some 12 years since I read it, but it very much talks about getting Jeff'd - and about avoiding it by abandoning the dream of creating mainstream products with million dollar potential. Find a niche that is financially viable for you, but isn't viable for Jeffrey.
Anyway, excellent talk. Went from "why am I watching a talk about languages in my spare time" to "I'm going to recommend this to everyone IT adjacent" really fast.
You're talking about "Why the niches is the name of game " chapter in SsSs, Start Small Stay Small book. It's one of my favorite Software Business books, but it's about creating software business not about Programming Languages as a business.
So this talk and presentation by Evan is a rare work about Economies related to Programming Languages; yet he missed to answer his own question, a big questions of: "How to make a living out of creating PL?"
Incredible timing that as the Google and Amazon antitrust suits get into gear, we get this really unique perspective into some of the side effects of their monopolies on tech, which at surface level seems unrelated
Insightful, blisteringly funny, and heartbreaking talk. Always a good time when Evan walks up to the podium.
Zig is a good example of a Donation based software project.
Wa going to say! Sometimes I get a little frusterated when it seems like news on zig is slow, but it's a small f/t team, backed by tonnes of overexerting contributors that have their own lives - probably a lot of burnout in that group. Still super entranced by the language and the design decisions
Evan is such a good speaker and presenter of his ideas. And a very nice and thoughtful person IRL from my few interactions with him. Elm is so lovely to build things with, but it got ground down by a combination of slow iteration and lack of flexibility (without escaping into JS). I think the language became 90% what Evan wanted, 8% what NoRedInk begged for, and 2% what the users needed. That worked well enough for NRI, but not a broad based community.
I still go back to Elm when I want to remember how much I used to enjoy client side development. And I wish Evan well - I’d love to see this project he has on his computer 😉
The "internet landlords" problem is explored by the economist Yanis Varoufakis on his most recent book entitled Techofeudalism. I'm still not sure if I'm convinced by his argument but it's definitely worth a read.
also - and more poignant - by Cory Doctorow
@@yeetyeet7070thanks I didn't know this name!
@@yeetyeet7070didn't know this guy thanks
Varoufakis talks about vassal relations which is a bit far-fetched, it doesn't need to be legacy terminology with the prefix "techno", all big corp liege to their own financial capital. Most futurists describe mega corporations as growing big enough to be governments of it own and virtualized citizenship by work attendance. You can easily think of Japan-like lifelong jobs in a single company.
What Evan talks more about is that "landlord" is more like a conduit and it can be dethroned quickly, i.e IE story, then Mozilla. Their own desire to gasp bigger market share can easily provide resources for 30 developers without flinching if that provides them with 40B in revenues.
Great talk! In the case of Linux, it is remarkable that Linus still holds so much influence over the project considering that anyone could have forked it along the way. But if anyone did it, the terms of the GPL would have forced them to open source it as well. Any patch that adds value in the competing fork, would have simply found its way back into the main fork. Is that the secret sauce? So perhaps the type of open source license ends up being a factor here if you want to avoid being Jeff'd. Maybe Corporate Alliance with a well chosen open source license ends up being the best path. Today, most devs contributing to the Linux kernel are no longer volunteers. They are devs working for corporations.
The GPL terms affect distribution, Linux has forks in every git(1) tree, but it's simply less work (more economic) to share the changes by including them upstream because Linus's tree is the key source and distribution.
Linus has taken breaks, there are lieutenants, but a corporate trying to dominate will be doomed by backlash of other corporates and communities. It's in everyone's interest to work with Linus's tree, but there are others of significance like linux-next and linux-mm.
Unfortunately licences giving more leeway have gained traction which allowed embrace and extend and fragmentation.
The FOSS model works best when many parties have an interest, big companies are big users and big creators of it.
@@RobBCactive why did you write "git(1)"? I'm guessing this is `man` syntax but even so I don't understand why you chose to use it here.
Ideally, every time a financial transaction occurs on the internet, a portion of the revenue should automatically go to the open source projects that the transaction depends on.
A talk on programming, economics and Georgist theory delivered with Nathan Fielder style dry humor, I'm truly terrified by how well the UA-cam algorithm knows my taste
So nice to hear from Evan, I learnt Elm when I wanted to share my programming with more people than those with python interpreters, I could never understand why people would want to use 1000's of npm packages over it. I really appreciate hearing his perspective as it's not something I would have thought about myself. Thanks mate!
Strange Loop talks never fail to blow me away
This doesn't apply just to languages, is about any piece of software you want to build and hopefully sell. Very interesting, insightful and fun.
one of the best tech talks I have watched in a long time.
Slight correction: economists and other academics don't *need* to use STATA / Mathematica. Many of them are switching to R or Python, it's just that licensed products have been part of the curriculum for too long.
Correct. I'm in the sciences and I'm seeing a big migration of all our stuff away from Matlab to python.
@@Joemakatozi1776 Yeah i took statistics last year(finished my CS degree now) and apparently only that semester had they finally swapped over to Python from Matlab.
This talk is based af. There's so many factors that I hadn't realised before. Respect for speaking on such a complex topic so candidly.
One "Design Incentive" was missed, at 30:23, which is a generalization of "Editor Licences" . This generalization is simply "Product", i.e. building a product jointly with the language design&implementation. There are three examples of programming languages that I know of, that provide a product:
1. Odin: their product is EmberGen/JangaFX.
2. Jai: Jon Blow is designing/implementing the language jointly with the game his company is working on.
3. Mojo: the most recent language that fits in this "Product" bucket, due to their "Inference platform". Mojo is additionally within the "Hosting" bucket (as they provide their hosting as well as supporting other cloud hosting providers such as AWS). Similar to Julia, I can imagine they would also go into Consulting or Research Grants (not PL research, but rather Machine Learning/ML & CV research)
Personally I believe this is the best way to design and implement a language. Create something that has proven to provide value.
Solidarity and empathy.
Loved the slides and the information flow.
I like his sense of humor. Interesting talk.
An independent langauge that does the donation principle is Zig
One of the best talks I've seen this year!
Excellent talk. Really appreciate you sharing your insight. Hopefully we can all get a little less Jeffed over time.
I love how the slides are all on this one giant canvas and can be zoomed in and out of. Does anyone know what software he might be using for this presentation?
Prezi
Thank you
Beautiful and insightful talk. Evan always makes me wonder what could've been if cs wasn't plagued by capitalism and clout seeking monkeys.
Google needs firefox to stay alive so they dont get accused of being a monopoly
dBASE II was developed in the old days by Wayne Ratliff to help him win the office football pool. A really effective DATA-CENTRIC language and turned super-successful in its heyday. I find it odd that most people use MEMORY-CENTRIC languages today for database project. Definitely under-optimal. I wish it would come back to DATA-CENTRIC again.
great talk!!!!
There are a few languages other than the mathematics/academic oriented ones that use the usage license model. They're kind of obscure though and really the usage isn't for the "language" but for the compiler/interpreter. K is one that comes to mind. There are a few Lisp compilers that also use this model. And a number of the old school business languages like FORTRAN and COBOL have compilers like this too.
There's quite a few that sprung up to fill the void created when MS killed Visual Basic 6.0. Such as PureBasic, RealBasic, etc.. they're niche indeed
Would be interesting to extend this to frameworks and libraries like React. As a web developer I don’t have much choice in what language to use but there are big companies and economic interests behind these libraries that honestly I haven’t given much thought to until I saw this talk.
very neat talk. much appreciated
Love this dude
I’m curious how Elixir has managed to do relatively well lately, does anyone know much about their history in this sense?
My bet would be on the Erlang (telco) legacy and BEAM being a masterpiece in what it is good at.
Erlang/OTO (BEAM)supremacy in concurrency + Ruby syntax(and many Ruby lovers moving over) + Good ecosystem + Pragmatic but thoughtful leadership + Phoenix(productive and powerful web framework)
A great book that helps with strategic thinking for your software project - “Inside the Tornado”
When you’re bowling you line up your shot so that the first pin hits the others in a chain reaction.
When you’re doing a software project, you line up your work in a similar way to cascade into growth/features/community/etc.
Unison could be an example of "Hosting" language
Someone give this guy $400m
Very thought provoking
it is the best talk for me,
so valuable, so deep, so generous.
Sorry to ask a question unrelated to the content of the talk (just a few minutes in), but how are the slides made? The half-page slide thing looks stunning.
Prezi classic 😊
Likely Prezi, he used to work for them.
This is the new Prezi version
While it's likely Prezi, because that's who Evan used to work for, you can achieve the same result with PowerPoint.
This is excellent. Thank you so much. It seems that there may be a relationship between Conway's Law and the points you make in the presentation. It seems that patronage is the way out if we can get individual developers to contribute both time and money to some sort of "Code Co-Op".
Incredible, thanks!
Once Evan starts talking about Traffic Acquisition Costs my brain immediately goes to "economic rent"
Thought-provoking talk. Well done.
If the developers could capture even a fraction of a fraction of all that rent revenue, without having to ask corporations... I think that would be a more ideal situation.
Would love to see a conversation on this btwn him and andrew kelley of zig
Good stuff
This is the exact problem I’ve been thinking about for a long time now. I have a project that I want to make my “the” project. Build it, plant it, socialize it, inspire someone, give my thoughts a life and watch it grow a forest. But I simply don’t know where to start. Everything and everyone is a propaganda. Do I just choose the lesser of the two evils and start?
Yes
Wait, did Evan just say he made a lamdera?
Well delivered
That google revenue reminds me how much I would like to use Yacy for searching.
Would having a "cooperative" contributor license solve this? A project is open source, might seem very interesting to some, but they don't know if their work will ever get recognized. If your project would use a "cooperative license" then the project success would be shared with all the contributors relative to their inputs (merged PRs let's say).
It depends what you'd define as "success" to share, no?
If I understand nothing else from this, I truly respect and appreciate you just from @40:16
Mandating open standards and marketplaces would go a long way to reduce switching costs. The economic costs of allowing tech monopolies is probably incalculable.
Incalculable but also unnoticed by most. The uninvolved would say "but we want Windows", nevermind that they didn't actually use windows in Windows and would use OpenOffice on a Linux box without even realising as really they were Word users.
Now Apple or tablets don't have the issue because users don't perceive a FOMO, they just use that cute little device.
"Power users" will however invest time to find problems to protect their time investment any change in accustomed work flow will be vehemently resisted.
Open standards and market places are a public good but don't have the highly motivated financial interests seeking to monetize something.
By "hosting" does he mean maintaining build servers and VM servers for people to use? Like being a cloud hosting provider devoted to a specific programming language?
09:15 Wow... imagine taxing advertising proportional to how annoying it is?!?!
ua-cam.com/video/XZ3w_jec1v8/v-deo.html
28:22 In the matrix of successful languages, certainly Java, Ruby and PHP are glaring omissions. It's interesting to consider where they fit -- and whether they don't, so the classification doesn't quite work.
As for Dart, Brendan Eich once bagged on it as a "Lars Bak retention program." Bak left Google in 2017 -- by which time Dart had been broadly dismissed as a failed "attack on web standards."
The correct way to understand Dart, I think, has always been a clean-room rewrite of Java. General-purpose, simple to learn, "functional enough." Static types and parallelism that stay out of your way -- and are thoughtfully considered, rather than bolted on. Cross-platform compilers that are not experimental, hobbies, or vaporware. The best tooling, period. A proper language spec.
That's what 30 engineers are for. If you're using a language from the 1990's with 5 engineers trying to roll a boulder uphill, give Dart a try.
What a great presentation!
Mozilla comes from mosaic killer and not a strange mispronunciation of the delicious cheese?? That… makes a lot more sense…
When the Language Author begins his own financial detective journey/forensics...
Key take away: 29:35 YOU can get Jeff'ed by EVERYBODY.
At 20:33 what does he mean by the "switching cost... for small authors"? Is he referring to the cost of switching TO the independent language? And how does this relate to get Jeff'd? (I.e. getting your users taken by a bigger company which is hosting your language without having to do any compiler engineering)
Miguel de Icaza sold his Mono C# runtime to Microsoft so he found a way to make money with his runtime. (Though he added to it cross platform app development in C# for IOS and Android and this is probably where the money was, Not in the runtime directly but in everything around it).
The internet landlord idea matches pretty well what Cory Doctorow is writing.
There should be no landlords in the internet. But developers should still get paid. And those paying them must get their money from somewhere.
I love the fog metaphor - Great way to start Evan! 🙂
Great talk~ Really enjoy it.
Í understand. I am a salesforce (cloud only) consultant. I did something no other consultant (I know of) can do. I can make salesforce code (apex) run locally. And I don't know what to do with that (kind of). Dont want to get Jeffd.
Lovely.
ZIG and Bun.js should be in this by default Open Sources; both being new Languages, so this is succinct enlightenment, thanks.
Bun isn't a language though. It's a JavaScript runtime
Can someone explain: on 4:47 he talks about Google's revenue and immediately splits it into expenses, does it imply Google spends exactly all of its revenue on these different things (like TA)? Aren't the revenue and expenses the two different numbers for any given company?
On the left, where everything is clumped, is their incoming revenue. The paths that split off are their costs. The path that continues straight ahead would be their profit.
There's a bunch of numbers used for different purposes, but the split shows expenses (operating expenses, other costs, traffic acquisition costs) and profits (operating income). Once you subtract taxes from operating income, and then subtract non-operating expenses (e.g. interest on bank loans) from that, you get "net income" which is the bottom-line of what the company made.
fantastic talk
Similar dynamics exist for open source and AI projects, please let me know if anyone figures these economics out... or if theres more insight. Big AI players like huggingface exist on jeffing researchers for hosting (+consulting)... success Is more about peoples perceptions/distribution than advanced tech
Landlords are everywhere you look. You start seeing them everywhere. You buy a toaster and then it's yours, but I know you have all kinds of stuff that you pay for that isn't really yours because everyone does.
So, programming languages as a service : PLAS
Exactly.
Very funny and frighteningly inside full 😀
I'm big Elm fan, it's very cool language and exploring it I got much better in multiple other areas. I really think there is too much overthinking about being Jeffed here, when you touched about consulting you mention reputation plays a role, the same goes about hosting, people would just use author's service if it's half decent just to support the language and author, this is something that can't be Jeffed from you Even.
So what about Java? I was waiting for it to be mentioned but never got to see the category for it...
What a fascinating and slightly tragic talk
When you look at Firefox revenue, keep in mind that the company that makes matlab has higher revenue.
I really loved Elm... you should've let people help you when they were offering... sharing is caring....
Evan I appreciate you, and I hope oneday we escape the consistent ugly reality of creators and builders getting Jeff'd. there are ways, but they're still in their infancy more than a decade after their first big adoption wave.
Can someone define hosting?
Google, this rich librarian.. damn !!!
Maybe is time to touch grass and forget about computers. We will be forced to anyway!
not for nothing: everyone should learn the ideas in Capital by Karl Marx. programmers aren't immune whatsoever to politics, and almost all of the effort we're currently spending on this discipline is being wasted by capitalism. we're spinning our tyres and we're beholden to politics to fix this. i wish i could bury my head in the sand but it doesn't work.
You forgot some of the most popular programming languages in your categorization - C, C++ and Java.
amazing talk lol
AGPL?
00:09 Evan designed the programming language Elm over 10 years and wants to share what he learned.
02:13 Languages are funded through corporate and independent sources
06:40 Google earns significant revenue from sponsored links in search results
09:01 The author discusses wealth and poverty disparities and proposes a tax on rent.
13:30 Open source software enables free development and distribution of software.
15:39 The birth of open source resolved the conflict between platform developers and app developers.
19:24 Developing a programming language is costly and puts small authors at a disadvantage.
21:16 Google's market dominance and how it impacts competition
25:02 Focus on the language, consulting, research grants, and editor licenses are important for success
26:53 Language licensing and switching cost are key considerations for commercial usage.
30:41 Different incentives exist for different environments and goals.
32:27 Language features depend on context and purpose
35:57 Donations and hosting provide different samples of users.
37:50 Hosting Elm code on Postgres with custom types.
41:41 Language design varies based on the influence and needs of corporations and independent developers.
43:20 Address switching costs for small authors
Crafted by Merlin AI.
No, the term "open source" did not come from Mozilla. The Mozilla release actually happened before "open source" achieved currency, more or less concurrently with me and a few other people deciding that "free software" needed to be rebranded in order to go mainstream. The rebranding took place in the year following the Mozilla release and got applied to Mozilla retrospectively.
You can put an anti-jeffing clause in your license
Methinks I like this talk
C++ is too expensive.
The landlord collecting rent on all of us might not be bad, as long as we have some democratic control of where those resources go.
"you would jeff me"??!
Beautiful.
soooo funny and insightful
Awesome talk! very insightful. Thanks Evan!