<!everyone> I've decided that what this community needs to really take off is just a little more resources. To this end, I'm raising $20M to fund this community. This is no joke. I asked Alan how much we need for the real computer revolution and that's what he said. Let's go do it!

Alan Kay wants us to have money! 🤣

What are you planning to do with $20M?

Wow

Give it to y'all of course!

^^

Exactly!

How come $20M? I’m not rich, but $20M doesn’t sound like a lot of money compared to what goes on commercially in what Alan Kay would call the “pop culture” of Silicon Valley.

Maybe a few people with projects already underway could use more help now with much smaller asks

How will you choose how to distribute it? Can people / projects register their interest somewhere?

You'll have to ask Alan why that number

$20mil is a good amount for angel-style (where 2-3 people get 1-2 years to get something off the ground)

(this is all provisional of course. Alan and I are just brainstorming.)

Has the Viewpoints institute received less than the $20M Alan says is required for the real computer revolution?

Well, even having a small slice of that amount would be hugely valuable and allow me to focus on FoC work full time.

it’s about 1/10 of what Eve managed to raise I think

You need anything from the community? For example, where the $20 mil is coming from?

Great question, Don! Hold tight for the moment. Still ironing out the details but will keep y'all apprised every step of the way

Lol, that's funny ... This is no joke. Oh sh... this is actually happening? Wow, I love this! I was just about to ask if anyone knew of foundations/grants to projects like this, but if it doesn't exist, why not create it? Amazing 😄 !

This is awesome! Planning on doing it worldwide or just US?

@stevekrouse Good luck =)…

Also, 20 M seems good enough. 500k per project might not be enough to sustain it in the long run, especially in US/UK, but it’s enough to get at least a year of development/raising

Anyhow it will be raised and used, it is a really generous step for the future of coding!

Worldwide!!!

Even in France? 😉

Thanks, @stevekrouse. I think I could put a million or two to good use. Put me down for that much?

No way! Are you serious?!

Oh, no, sorry, I mean I could spend a million or two.

Hahahahaha oh ok

Sorry to get your hopes up.

Tell us more!

@Peter van Hardenberg you mean like shorter R&D sprints on a topic? (https://www.inkandswitch.com/archive/rdlab-concept-sketch.pdf)

If anyone here wants to volunteer donations or investments (we're accepting both), feel free. Of course no pressure. I'm sure I'll raise the funds elsewhere

Adam's position would be substantially more definitive as the creator of the system, but I'd say that the Ink & Switch model creates alternating modes of intense, cross disciplinary collaboration with writing and thinking.

I'd be curious to explore what the relationship between this slack and the Ink&Switch Slack is. There's so much good stuff in that Slack I wish were automatically posted here (and some vice versa)

I don't want to imply that this model is necessarily off-the-shelf useful to what @stevekrouse is proposing, but we would entertain project proposals from any of the standing lab members that included research goals, staffing/skill needs, and a budget estimate. These proposals were the result of what we called "pre-infusions" where a lab member or members would try and generate several research options through a mix of preliminary work and survey.

Good Luck! And please start a YT channel! 🙂

YT?

YouTube, because that sounds so fun! 🎉 🙂

Oh good idea

I think a lot of young folks (maybe not experts) who are hunting for tutorials and better and better solutions are out there on youtube. Their talent (or at least attention) might be useful. To generate this attention in general and to show the fact "its happening" it would be nice to see a progression like on this channel: https://m.youtube.com/channel/UCbfYPyITQ-7l4upoX8nvctg On the other hand as far as I have seen Alan might think the top inventions came from the same group of people, but still, accessible videos are a low-barrier entry point to engage.

@Peter van Hardenberg what’s the status of ink & switch?

Great idea Pezo! Do you know anyone who would be good at creating such a channel?

@ibdknox quiet, at the moment. some folks are spinning out work we did at the lab into companies and we have a few smaller-scale on-going things

@stevekrouse Not sure, but I am quite sure that the folks who built large audiences (100K-800K subscribers) out there (just google for “youtube programming channel”) would be happy to share this fact (as a piece of news) if something is going to happen for real.

Well said, I agree

Great! I was a bit afraid even to mention! 🙂

Are we imagining a coordinated effort? Prizes? What have the roadblocks been? Why hasn't our real computer revolution happened yet? We've hard plenty of time. Perhaps we need a counterbalance to some bad equilibrium? The walled appification of things?

Why hasn’t our real computer revolution happened yet

Well, for one, I haven’t had much free time lately… 😬

Deploying capital effectively is exceedingly difficult, I'll be curious to see how you end up doing it 🙂

You and me both brother

Realistically, I think we needed about double the time we had with Eve, but the hard part is that I think we also needed the time away we've now had. I'm sure we would've landed somewhere interesting had we just kept going, but we've learned a lot taking a break from the problem that I don't think we would've otherwise.

You can always try again! I consider Eve a success, the influence it had will surely leave a mark on our future

I also think Eve was successful, just not in the ways we thought it would be 🙂 e.g. one of the most interesting things that came out of the project was the style of working that we developed

@stevekrouse “You’ll have to ask Alan why that number” - We’re not worthee! 😀

can we get an invite link to the ink & switch slack and any other interesting groups people are a part of? i think the first step here would be consolidating or at least federating the ad-hoc communities in the interest area

Over on #CJT25RWKE, we've been discussing the idea of getting a group of FoC researchers under one roof. Daily face-to-face conversions between FoC'ers with different perspectives and expertise could really accelerate this thing. A major roadblock to making this work is funding.

The piece of mind perspective about funding: I'd rather have $500 or $1000 a month for many years, than many thousands dollars per project that I'd have to constantly hunt for.

the style of working we developed

@ibdknoxcan you say a little more or point me to link on this

@stevekrouse I'd be happy to help your fundraising process. by that I mean, I'd help build a database of all potential funders, and help build some tools to automate some of the grant writing process. can also make a lot of introductions. i think it's a good idea

I admit that I checked twice to be sure it's not April 1st. But all my electronic devices say October 9, so that must be right 😃

@stevekrouse:

Alan and a committee will distribute the funds

Anyone on the committee probably can't also apply, so the committee will be people from outside this community. Might that mean that they are people who don't necessarily understand what is important here, etc?

@ibdknox Seconding @cogell's request to elaborate on the style of working y'all developed. (I had this comment drafted for a couple of hours.) Your comment about the value of taking time away from a (formal, funded?) project really resonates as well. That's one reason I tend to be suspicious of funding: it takes so much effort to get that when you do get it it's often not the best time to have it. Projects need more and less focus over time, and I don't know how to organically allow for that when dollars enter the picture. Which is why I work on my project on the side. But I'm probably erring too far in the other direction. I'm sure there are times when I would benefit from focused attention to Mu, and I can't do that, and I often am not even self-aware that that's what I need. (I echoed similar sentiments a couple of weeks ago: https://futureofcoding.slack.com/archives/C5T9GPWFL/p1568913088387500?thread_ts=1568900621.372600&amp;cid=C5T9GPWFL)

@Kartik Agaram I agree. Not only does applying for funding take time away from work (particular when multiplied by the number of applicants), it can be demoralizing not to be accepted and make people feel that if they didn’t get funded, their project may not be worth working on. If I were funding projects like these, I wouldn’t accept applications at all and just give to those who are actively doing what I considered to be promising work.

If I were funding projects like these, I wouldn’t accept applications at all and just give to those who are actively doing what I considered to be promising work

this is a good point...perhaps patreon or github sponsors should standin for "an apllication" process

I see where you’re coming from, but I disagree. Yes, funding does take aways time if you are actively looking for it and spending time doing it. But a short application process and a interview or two won’t take much time, and those who are interested can apply. Not everyone is looking for funding on a project, there are researchers that don’t have projects up on github because they’re doing it in a lab but might be doing amazing things, not all projects are open source and even a project being open source and having dozens of active maintainers doesn’t mean that they will want/need funding or know what to spend it on. For all of those reasons I believe it’s better to accept applications a la YC, so people who want funding for their projects can apply and that doesn’t stop anyone from contacting those who do interesting stuff and are outside of this community and inviting them to apply.

Wow, if this goes thru, it could be a significant push for this area of research..

Actually who needs $20m? $2m would fund 10 of the one-man projects listed in the projects spreadsheet for a couple of years to get a production-quality MVP out.

This is super cool! Have you checked out the NSF grands at all? https://seedfund.nsf.gov/apply/

@Wouter has a great point. Although you can get a cool prototype out for a couple grand, to really make a dent in the universe takes a substantial amount of resources. For Tree Notation I put out plenty of prototypes and posts and a few papers but it wasn't until we had lots of core tooling that it started to get any traction at all...we've put a few hundred grand in direct costs into the project and opportunity cost of free labor could easily be in the millions. if you look at the FoC projects gaining traction (things such as Observable, Rust, Lean, etc) they have millions at least in backing and resources...the YC partner Tim Brady said recently that the #1 mistakes their 30 non-profits make is relying too heavily on donations and not finding a sustainable business model....I look at HARC and think perhaps it might not be too helpful to get $20M....perhaps getting a smaller amount and finding a sustainable business model would be a better approach

500k per project might not be enough to sustain it in the long run, especially in US/UK, but it’s enough to get at least a year of development/raising

The median pre-tax income in the US is ~30k. So 500k is 16.6 person-years if you don't live like rich people. Or 32 years for one person if you use a standard retirement savings model. 600k gets you 86 years - exponential growth is not intuitive. $20m could fund 30-40 people to work on FoC for the rest of their working lives. That might be more interesting than giving all the money to bay area landlords.

@jamii make a list of all people on this slack, and where they already live.. and see how many of them could reasonably survive on 30k pre-tax.. I bet it is close to zero. Though if you are suggesting we should all relocate to a commune in Wyoming to make FoC happen, that's actually a pretty fun idea!

I assume the math works out similarly if you give a $20m endowment to a non-profit that funds 30-40 researchers each year.

as long as we’re talking utopian communes, we could drastically reduce living costs by putting a chunk of the money into rent-free real estate

@Wouter I live in a nice apartment between two parks in Hackney, the borough of London which includes Silicon Roundabout. The median salary here is $34206. I live quite lavishly on the equivalent of $37000 pre-tax, a large portion of which is spent on international travel. Reducing my income to $30000 would mostly mean taking slightly fewer climbing holidays in Greece. (source https://assets.publishing.service.gov.uk/government/uploads/system/uploads/attachment_data/file/791181/NS_Table_3_14_1617.xlsx)

also @stevekrouse on the youtube front, visual aids are super helpful for visual programming 👐

where on the scale between "VC investment" and "research grant" is this endeavor expected to fall, both in terms of application process, and outcome expectations?

The bay area is a disaster but even in New York, for single people $30k is quite reasonable. For 'non-family households' we have median $38446 in Brooklyn and $22,827 in the Bronx. https://factfinder.census.gov/bkmk/table/1.0/en/ACS/17_5YR/S1901/0600000US3604710022 https://factfinder.census.gov/bkmk/table/1.0/en/ACS/17_5YR/S1901/0600000US3600508510

FWIW, a $30k/yr position would not even register in my consciousness in a job listing, and if an offer like that came at the end of an interview process I would be offended. The reality is that software engineers cost six figures, and that they're concentrated in cities, especially SF. I think few engineers would be drawn away from their metropolitan centers with a 5-10x pay cut. VC investment sums reflect this cost— If it were possible to start a software company for 10% of the cost, I am sure investors would be doing it. Also I lived in Brooklyn for a summer two years ago, and it was a strain on a six figure salary. My costs were higher than where I am currently living now, in Berkeley. $30k in the NY metro area is poverty.

I was thinking more $30k/yr to work on your existing research project, no strings attached. Hell, call it $60k/yr, more than the median household income in almost every city in the US, more than the average postdoc, and you could still give 13 grants a year forever. https://academicpositions.com/career-advice/phd-professor-and-postdoc-salaries-in-the-united-states I've derailed the thread enough, so I'll leave it at that.

Well it seems that there are big disparities in income needs on this community. I am more on the Jamie's side (even if 30k$ is a bit low) and always astonished by the very high salaries needs for places like SF (but I am sure I will understand them as soon as I work there 😉 ). Sure @jamii is right, based on this simple fact, some projects will need more funding than others. Of course it is more complicated thant that (it depends on the project and on the people) but it is a parameter in the equation.

$30k/year after tax? That’s just a bit more than I earn now over here in Croatia (after tax) and considered “high salary” in the tech industry. Give me taxfree 30k and you got me covered for a year.

The last time I checked, a decent one bedroom apartment in SF was ~$4K/month. Looks like average salary is $90K/year https://www.payscale.com/research/US/Location=San-Francisco-CA/Salary

🙀

@Peter van Hardenberg How did Ink&Switch raise money? Can we do the same?

Yeah, cost of living (COL) differences are kind of insane: I think COL is a function of capital density and transit times, basically a number represented by the amount of capital you could reach in a given amount of time. It literally couldn't be cheaper to live in SF (unless you made transit faster/cheaper or increased the available space both of which take 10+years and is highly political since it changes power dynamics) [ About remote: I've worked at a large 100% remote company and there are clear tradeoffs compared with in-person companies (DM/thread elsewhere for more details) ]

Fwiw, I would happily take a hypothetical $30k salary to work on my FoC project; currently I have to earn that $30k doing other kinds of work. However, I'm not sure if basing the feasibility of a $30k salary on the cost of a San Fran apartment is sensible. That's more than enough to live off in Melbourne, Australia, which is considered a city with a high cost of living. Such a salary would mean I'm not going to have a luxury car or annual vacations to Europe, but I'm not working on my FoC project to generate wealth, I'm generating wealth so that I have the means to work on my project. I don't think I need to live in SF, but I DO think I need more connections with like-minded people. It's clear such people don't all congregate in SF; people in SF are busy optimizing ad clicks instead.

Haha I know it so @stevekrouse you are going to build a distributed organization? 🙂

@ibdknox have you written about the "way of working" for the Eve team? I'm extremely interested in other teams like yours that bridge an academic / industrial gap

@Peter van Hardenberg How did Ink&Switch raise money? Can we do the same?

@Duncan Cragg it was essentially pre-angel money with the hope that start-ups would eventually spin out (which is now happening at ~n=2)

There is (1) FoC for large scale industrial programming; (2) FoC for automating white-collar professional work: end-user programming for data scientists or physicists; and (3) FoC for Alan Kay's vision of Personal Dynamic Media, personal computing outside of a job context, live coding as an expressive artistic medium, FoC for kids. Does this fundraising project prioritize any of these goals over others, and is the same fundraising pitch applicable to all 3 kinds of FoC?

@Peter van Hardenberg @cogell @Kartik Agaram I haven't written anything, but I should at some point 🙂

spare us a few words to wet our whistles?

Sure, a couple of things: we believe that you start slow to go fast and so we spend a fair amount of time up front trying to determine all the ways in which whatever we think is true is wrong. Basically "pre-morteming" and exploring as much of the space as seems reasonable. We assume that we're wrong about pretty much everything, but that we can still be usefully wrong. To that end, when we start prototyping we acknowledge that we're at our least knowledgeable and plan to throw several versions away. From an engineering stand point, we've found that taking dependencies slows us and whatever we're building down by quite a bit, so we take very few if any. When we build something we try creating an end to end stubbed out demo and then fill it in over time. This allows us to see the system as whole from the very beginning, seeing potential issues at all the levels (UX since we can click through, performance bounds - the stubs will be the fastest it will ever be!, architectural weirdness, etc)

❤️ "very few if any dependencies". Then again, I'm probably batshit insane on this axis..

I'm pretty certain that's far more important than it might appear

The place where I worry I take a good idea past its breaking point (like socialism): minimizing dependency on existing languages. I've never met a language I liked, mostly because I include the compiler and toolchain when I evaluate them.

@ibdknox Doesn’t the idea of “architecting for performance” conflict with the goal of making throwaway prototypes to explore ideas?

nope, or rather, it doesn't have to

They forget to throw the first one away.

☝️ you're learning how to build something fast with each iteration

@ibdknox it’s really hard to google “hyper wikipedia single machine query”—results are the wikipedia entry for hypervisor etc. who are the hyper folks? what links or search terms should i be using?

bless you

But doesn’t a focus on performance compete with the primary goal of finding an accessible programming model? You need to keep the former in mind for sure, but perhaps not design explicitly for it? How slow can you “accidentally get” whilst you’re developing the programming model, assuming you’re still willing to throw away your prototypes?

I don't quite follow your question yet, I think. I don't see discovering what to build and how to build it as orthogonal processes.

i think that if you understand “accessible programming model” to mean “language or set of behaviors that communicate what you’d like to know or do” then it falls out that performance problems are languages, tools, etc that make it easier for you to ask the computer to do a lot of unnecessary work than it is to formulate the question you’re actually asking (sorry for the complicated sentence structure it’s 1am where i am)

Oh, just as a gut check - what I've been talking about is how we've approached building any system (most of which lately haven't been about programming at all, so the goal isn't to find an accessible programming model)

Ah, I was thinking about how that process applied to Eve. I guess I’m talking about the difference between developing the UX, and developing an implementation. Ideally, you can separate the former from the latter, so your preconceptions about what’s fast don’t creep into the user-facing parts.

Depends how out there you're going - in several cases that was impossible and led to dead ends for us.

chris please go on more podcasts or whatever

I’m nearing the point where I’m going to have to find that out for myself. Designing for both UX and performance simultaneously definitely adds an extra dimension to the problem. I already had a hunch that it was going to be necessary so I guess I’ll take your experience as a caution!

it’s always illuminating when you weigh in on this stuff—especially the talk about the different evolutions of eve as a startup. i think a big barrier to more progress on the FoC front is that it’s really hard to even get started really “going for it” and even if the VC model didn’t work out for eve a few years ago, experience reports about throwing yourself into it full time are lacking

I’m talking about the difference between developing the UX, and developing an implementation. Ideally, you can separate the former from the latter..

@Nick Smith I think this is the core of our disagreements in the past. To get metaphysical about it: when can you decompose a feedback loop and stop treating the systems you build as experiments showing you how to build systems? The answer is "never". Or at least, not for our lifetime. Software is still a young field. Permit me a tangent on waterfall. Hopefully I'll be able to bring it back on topic.. I've been through three levels of understanding of waterfall: Level 1: waterfall bad, iteration good. Level 2: even the original paper about waterfall treated it as a strawman: https://en.wikipedia.org/wiki/Waterfall_model#Royce's_final_model Level 3: actually reading the paper (http://www-scf.usc.edu/~csci201/lectures/Lecture11/royce1970.pdf) Waitaminnit, it's not just the first model that is waterfall. All the models in the paper are waterfall. Successive refinement of the waterfall model. Waterfall has feedback loops. It's not like water falling at all. More like a circulating fountain. Why is the strawman so seductive, why are the feedback loops so easy to forget? • Historically there's been incentives to forget them, driven from the top by people without design experience. This is the well-known reason. But I think it's not the only one. • People often learn how to design from an expert in a narrow domain. Maybe your first job was in a compiler shop, and you saw the master crank out compilers. And he pretty much flowed down from requirements gathering to productionization. What you didn't see was the first few compilers he built, when the feedback loops were extremely salient. Careers tend to follow an arc where people build similar things over and over again. And seeing someone later in their career (whom you may admire), it's easy to miss the feedback loops that used to exist but gradually grew vestigial as they over-trained on a single domain. Extrapolating from this little rant, it's seductive to linearize workflows because the world wants work to get done now, because of pressures from non-designers, and because staying in one domain linearizes with good reason. But when one seeks truth, the feedback loop always exists. Even if it hasn't been used in a while, we dismantle it at our long-term peril. Building the implementation must always be allowed to flow back and trigger a redesign of the UX. If you only learn from the UX step your exploration of the state space is crippled.

i’m half asleep now but hypnagogically moaning “think of all the code that doesn’t run in dynamicland when they use an electric fan to rotate a lazy susan representing a plastic volume dial in physical space instead of a react component, mouse driver, graphics engine, css, etc”

I'm not sure how this became a conversation about the waterfall model -- I was never advocating for any sequential process. I don't have any other strong opinions on this. If that clarifies things sufficiently, then I think it might be best to leave this thread to a discussion about the original topic (R&D funding).

tbh that sounds like the perfect way to work, similar from what I’m do/want to do. Especially when working on something that’s FoC related, most of dependencies weren’t meant by the developer to be used like that - I noticed how much work I had when I tried to use a generic “code editor” dependency and realized I’d be better off having to reimplement the whole text editor on my own again because it failed failed. Faster iterations really do allow you to write more performant code - When you get the system in your head with the 1st, you already know the edge cases you covered and can probably find a better way to handle it all easily for v2. And then in V2 you get to explore even more of the problem by having the “mental map” of the system already, so you can navigate the concepts more easier, merge them, discard them, find new ones. It’s like moving into a new city: At first, you’re lost and find a few places (store, barber, bakery, coffee shops) near to cover your basic needs. And sometimes you get lost in the bad part of the city trying to find them. But you notice they’re too far away and you know what you need now (and what part of the city to avoid) so you say screw that, go to explore and notice that you can get a better coffee in the shop 2 streets away and grab bread on the way back (maybe even in the same bakery, maybe not. at least you know it’s there and it’s an option), so now you go there and grab better coffee and bread faster instead of just sticking to V1, 🏃 running to the same bakery and coffee shop, spilling coffee all over you and saying “look at me, i’m optimizing!”

@Nick Smith it seems small compensation that you don't advocate for it, when you practice a sequential process in effect. You're just annoyed that you're worse than waterfall 😄 I think I'm very much on topic, but you do you.

This is such a great thread, @ibdknox. It really mirrors a lot of the working processes and thinking of the lab. One thing we've suspected is that taking VC money would constrain your ability to execute against the long time-lines required to do really ambitious work, and I'm curious whether in retrospect you found that significant? I also completely agree that if you don't design for performance at the beginning, you (probably) can't get it at the end. It's such a fundamental and formative constraint.

yeah, I agree and a lot of what I've learned about performance is from game folks 🙂

@ibdknox "3 people can do a lot more than a team of 100". That's because of communication overheads. In your experience, how much does it help if the 3 people work together in the same room, vs remote?

It's more than communication overhead I think. It's also the fact that teams have gravity - as they grow it becomes more and more necessary to do work just to support the weight of the team. They also think differently and start solving problems that tend to be ancillary.

20 million is a lot of money. If i recall correctly Eve raised 2.3 million from Andreesen Horowitz, and the Luna team just raised 2 million for their second round, and the Red team via an ICO raised an unknown amount, probably in the tens of millions. So 2 million gets you about a team of 6 people for 2-3 years, depends on the country of course. Red is pursuing a multi-platform initial release, so they have higher costs and a larger team. the best way to do this type of funding is to put at least 3 parallel competing teams on the same general project area, and whichever one looks the most promising after N months gets increased funding and the other teams are disbanded. This was the general methodology used so successfully under Thomas J. Watson Jr. who presided over the most successful and excellent tech company the world has ever seen. I know that most of you folks don't know how great IBM was, but they were amazing, and for a near monopoly, while Watson was in charge, the company was of the highest integrity and innovative like no other. After he retired it was ruined as a company, they became gutless and lazy. They are only a shadow of their former selves. But anyway back to this idea, by having competing teams you get peoples' juices going. A race is far more exciting than a country club atmosphere with no deadlines and a steady paycheck you won't get much. Look at how little gets accomplished in France. They are so smart, but so far behind in programming. I think it is just too comfortable. I had a friend David who tried working in Paris and couldn't stand the lack of motivation and quit his cushy job. If you are not going to have a VC breathing down your neck, at least put some danger into the process by having some competition. Some of the greatest creations in history were done very quickly. Like building the Pentagon (under 1 year), or the most impressive musical composition feat in a short time period in history, Donizetti's Don Pasquale which was done in just in a handful of weeks, which if you ever go to a big public library and see the score, you would be hard pressed to copy by hand that score in the time he took to write it. Nobody has ever done an opera of that quality in what amounted to just a few days. @stevekrouse , i suggest you study the biography of Thomas J. Watson, as he directed some of the most productive R&D, and was quite a forceful personality without being a chair thrower, etc. Hewlett & Packard were also superb tech managers and developers; actually very few great leaders in computer history. If you just throw money into a project without a strong plan, you get poor results. Take a gander at the doomed california high speed rail project for an example of how not to run a project. They spent billions, and accomplished nothing useful. Or the hot fusion projects of the last 30 years, which have promised fusion in 10 years for the last 30 years straight. Many R&D projects fail, so I thing Watson had the most practical approach to solving an important problem.

@ibdknox Could you please expand on “stubbed out demo”? How do you decide which parts to stub out? What is most minimal demo do you start with?

When we build something we try creating an end to end stubbed out demo and then fill it in over time.

An idea from a discussion I had yesterday with different people on a different topic: One way to usefully spend money on researchy topics with little overhead is organizing meetings/workshops/... and pay for all the participants' travel costs. It's a small-perturbation thing for everyone (nobody moves to a different job, for example), and therefore creates no long-term seconday effect ("what will I do when my grant runs out?"). But it's a big boon for those of us who somehow manage to work on our FoC projects but couldn't afford spending much money on it.

@Edward de Jong / Beads Project I like the IBM "build 3 and pick the best" approach. I think perhaps you could have "problem teams", and "solution teams". The problem teams would go out and define what the problems are, and build datasets that all the solution teams could use (like @stevekrouse’s recent dataset on next gen IDEs). That work is a lot more straightforward (but less sexy and underfunded). Then you have the competition on the solution teams. Perhaps, i dunno.

At the end of the day you are looking for some combination of 1) ease of learning and use during development, 2) ability to transfer projects from original author to maintainer gracefully, 3) aesthetic quality of the products produced, 4) target platform range, and 5) suitability for large data inputs. I don't think the future of computing is going to target niche markets like ML or exascale computing; those are best left to specialized toolchains custom made for that sort of thing. You could start by giving a product to clone, saying make me facebook in 3000 man hours, or pick some popular game, etc. Jonathan Blow's very clever Jai programming language is custom built around making his next game. The language does exactly what he needs for that game. And since it is a complex game, the language is pretty complex, and fairly low level as he is shooting for low power consoles like the Nintendo switch.

It seems like the OpenAI project has made an impressive amount of progress. If you’re looking for an example of how to go about this, maybe that project would be a good place to start, though the problem and solution space of AI may be much better defined at this point than FoC.

It's tough to use OpenAI as a measuring stick. The amount of funding and prestige that they had from the very beginning is incomparable to what we're talking about here. They paid a lot to get some of the best very early on.

@ibdknox Yeah, the level of resources is a big difference. Also, AI has converged on Deep Learning connectionism as the path forward, so it’s much easier see how to usefully invest in it. I recently watched a video of Danny Hillis talking about the Connection Machine and how it was designed for connectionism and essentially had the same computing model as modern GPU/TPU but a about a million times slower than what was required for connectionist networks to start becoming particularly useful. Right model, wrong time. Both supercomputing and AI had “winters” between their early days of promise until the tech got powerful & cheap enough to make them widely useful (I’m counting cloud computing as modern supercomputing). Lately, I’ve been wondering if the FoC winter between the early days of LISP and Smalltalk and today are part of a somewhat similar pattern. If so, maybe we’ll need a convergence of hw power, a sufficiently suitable sw stack and standards, to build on top of before we start to see solutions that can clearly point the way forward.

I share this vision of FoC (or programming) winter, and the need of a "sufficiently suitable sw stack and standards". And my opinion is that we know have a family of standards that can be used to ease adoption of new tools/languages/etc., they are the web standards. They provide all the basic stuff (or nearly all?) to build these tools/languages/etc. For me several years ago, there was one critical missing piece in the web landscape: a way to execute fast processings, that's when WebAssembly came in and that was why I was so enthousiast about it (and asm.js before it). It was the missing piece to build new stuff for massive adoption. For me, the next step to "exit the winter", is to build something on top of web standards that eases any act of programming, something that permits anyone to propose new ways of programming. A kind of integrated "FoC development" environment.

There are many next-gen language projects that emit JS and other languages so as to leverage the operating system layers (and it is clear that the web has slowly morphed into an operating system, albeit a very restrictive one). But the web is not particularly superior to Win32 or OSX or the previous OS'es. It just happens to have a billion+ user base, which creates an opportunity through the massive audience. But i don't see much running on the web that couldn't have been built 20 years ago. And don't think that WebAsm is really that big of a deal; the V8 engine is so blasted clever and fast, that only in the heaviest of applications can you tell the difference. I did some benchmarks a while back, and JS trounced Python for performance. So JS is already quite good as an assembly language which is how i view it. Where WebAsm will have a big impact is breaking the monopoly of JS which is not a particularly good language.

My point was that, as you say, the web "has a billion+ user base", and that's why I think building things on web standards is good for adoption. It is the major reason for me to build stuff on it.

@Edward de Jong / Beads Project “the V8 engine is so blasted clever and fast, that only in the heaviest of applications can you tell the difference.“ True, and the DOM layout and high quality text rendering is incredibly fast and can use WebGL when needed. It supports OS and language level sand boxing, gigs of client side storage, all sorts of devices like touch and gamepads, and good debuggers that connect editors and browsers. Even full sockets and mDNS are in the works. With tens of millions being spent a year on making it better, what could compete with this platform for building FoC environments on top of?

I see two big weaknesses in the web platform from a FoC point of view: 1) the technology and the culture behind it are ver much fashion-drive, 2) user interfaces are constrained by the page-in-browser-tab model.

@Konrad Hinsen 1) abstractions can separate one from this 2) check out Progressive Web Apps

@Steve Dekorte I am not a fan of web development tools. i am none too happy about big brother spying on my every keystroke, and the lack of privacy and security for valuable IP i could never tolerate in a business. The fact that the browsers are controlled by a few entrenched competitors, and due to the baroque hypercomplex and messy design of the web which just keeps tacking on more function like the Winchester Mystery House is not ideal. That these big organizations are immune to suggestions and changes, and bugs don't get fixed promptly, makes we yearn for a simpler less messy underpinnings. The web browsers refuse to let us use UDP packets, except when they deem it okay, and UDP is the way to go for high performance; so much less overhead than TCP. Google & its co-conspirators have been steering the web browser towards being an operating system bigger than windows. It was a clever strategy and clearly has convinced young people that it is the only way forward. I would prefer a real operating system, that is lean and mean, and not such a mess. CSS is one of the worst layout methodologies ever invented. Like the original US television system which was called NTSC (later nicknamed never twice the same color), i wish i could find an ancronym for CSS that involves mysterious movement of your blocks in baffling ways.

@Edward de Jong / Beads Project “i am none too happy about big brother spying on my every keystroke” I’m concerned about that as well, but there’s no requirement that a web browser be used with a server. I’m using single page Javascript apps straight from a local file the desktop. Also, being able to choose to drop the same file onto a server where everyone could run it at the click of a link is extremely valuable (consider how many fewer potential users would be willing to download and install an app vs clicking on a link). As for UDP/TCP, all the browsers will be supporting full UDP & TCP sockets and mDNS soon. Take a look at the Chrome or Firefox dev builds. There’s also no need to use either HTML (besides a few lines to boot your JS code with) or CSS (aside from setting style attributes on elements via JS). I’d agree that the DOM is not the layout engine I’d design, but it’s very fast and full featured, and runs everywhere. While it makes may things that used to be easy, more difficult (view management), it also makes many things that used to be very hard, very easy (text layout). The value of the browser’s sandbox is also worth considering. For security reasons, I don’t run any apps that aren’t from highly trusted sources on my trusted laptop anymore, and certainly no code downloaded from github. Maybe that will change once the desktop apps and terminal are fully sandboxed but it’s looking like that will probably come in the form of app signing from big vendors like MS, Google and Apple and I don’t want any of them controlling what apps I can run. The browser is the only escape (for typical users) from the app signing Dystopia.

@Steve Dekorte Nothing fast about the DOM. Jobs at Apple wanted Display Postscript, but that doesn't animate well, and the DOM is a poor version of a retained graphics model. I think they could have picked a well thought out graphics foundation at the beginning, but chose instead to constantly add more and more function to HTML/CSS making it the mess that it is. SVG shows the same creeping complexity, and who knows how many decades before the browsers implement all of the features. The fact that the feature set is invented and then a decade rolls by while the features are gradually rolled out, and with the freezes in the mobile world where devices are cut off from update streams, it makes for a bizarre situation where 89% of the user base has some key feature. What is a programmer to do then? It is chaos the way they did it. In the good old Win32 and Quickdraw days of Windows and Apple, you had a fixed specification, and a very solid underpinning. Software developers loathe this flabby "almost-implemented" state of the web, and it makes for intensely flaky software.

@Konrad Hinsen Can you elaborate on "the technology and the culture behind it are ver much fashion-drive"? Especially, I am not sure I understand properlly "fashion-drive" and what it implies in your view. But If I guess well, I suppose I second the point of @Steve Dekorte: "abstractions can separate one from this". When I talked about web standards, I talked about the technologies, not the people around it.

@ogadaki @Steve Dekorte You can always add a layer of abstraction, but that’s one more layer to maintain. Ideally, for a project running for ten years, you want dependencies to be stable over ten years. Can you imagine working with 2009’s Web tech today?

(developers: @Konrad Hinsen was just giving an example, please don't actually redefine any of my keystrokes or mouse actions, thanks)

@Konrad Hinsen If your concern is a project running for a long time, I can still run the first web page I made in the 1990s and it works the same and everyone can still use it, even on platforms that didn’t exist at the time like smartphones. Can you still run 32bit desktop apps compiled for whatever version of the OS you were using in the 1990s? Many of the OSes I used at the time (Irix, HP/UX, SunOS, NeXTstep) have been discontinued. If you want to reinvent a sandboxed, cross platform environment that’s open, crazy fast, has advanced text rendering, and runs everywhere, that’s cool - it’s something I’d like to see done and I’d definitely take a look at it if/when it’s ready, but until then I’ll use the one that already exists.

I'll just throw this link here: http://canonical.org/~kragen/eotf

@Kartik Agaram Or there's CollapseOS, for an alternative vision of post civilization collapse computing (collapseos.org). But the Long Now Foundation is my go-to for this sort of long term thinking. Kragen considers an artifact that is durable and reliable enough to stay in working order for an extremely long time, without maintenance. Consider instead a very long lived institution that preserves and maintains the technology. Like the European monasteries that preserved knowledge during the dark ages, the Church that ran 200 year cathedral building programs, or the Ise Shrine in Japan, which has been rebuilt every 20 years in adjacent sites for over 1,400 years. https://blog.longnow.org/02019/09/11/long-term-building-in-japan/

There are many aspects to computing over longer time scales. One is preserving a full system, as in the link given by @Kartik Agaram. A similar idea but for software only is proposed in https://doi.org/10.1145/2814228.2814250. Keeping archived apps or Web pages usable (as referred to by @Steve Dekorte) is a much more widely shared concern. What I referred to is yet something else: continued development over many years. Archived tools are not sufficient for that. As development goes on, you want to add new developers, new dependencies, perhaps new development tools. This requires a whole ecosystem and a whole community that values stability. Real-life example are Java/JVM, Emacs, or Common Lisp. In either one you can pick up code that was last modified five years ago and continue to work on it without feeling like you are doing archeology.

Quick overall funding update: I've gotten a lot of discouragement that I will be able to find non-profit funding sources. It seems possible, but a lot of work to find such visionary people who want to give us millions. Of course I'm open to this still... I've been encouraged to go the for-profit route. Much quicker and easier to raise the money. Of course the catch is that we need to give a return on investment. Luckily, the biggest companies of all time have come from this kind long term research (Ford, Bell, Microsoft, Apple), so it seems reasonable to be able to capture some of the value we create as long as we're operating at longer-term timescales than business as usual. I'm brainstorming ideas with Alan, and friends and family, all over the map: VC firm, incubator, studio model, single startup, etc. To be very clear: my top priority has always been the research to produce better media for humanity. Achieving profitability is just a possibile strategy to fund the research. So I will be very careful to shield researchers that want to do pure research from the commercialization side. In short: I am still working on this full time. I'm still optimistic. But this will likely take a while and I have no idea what shape it will take...

Well with the contacts we can all summon in this community I'm sure it's possible

(Sam Altman tried to get non profit dollars for HARC and OpenAI and failed. I'm sure it's still possible but seems hard.)

You may need to gather a small crowd of delegees to help you

Thanks a lot for the news on the progress! 🙂

Trying to raise $20M from VCs out of the gate is going to be tough. You typically raise successively larger amounts at certain proof points. Maybe if you're pitching yourself as a VC fund the math changes, but my understanding is that's a very tough sell as well.

Yes, and in addition, they probably want to be sold one (product) vision, not many

Yeah, it also changes the timing and incentives significantly

(random non serious thoughts..) so if you get a vc funded commercially successful product, from the payoff you can get the 20m for the real deal..

There are multiple well funded, research oriented non-profits in Waterloo ON CA. Perimeter Institute is the most famous. So the conditions for success exist here, as well as other cities. It's the best model for long term sustainability, but takes a lot of time and work to set up.

I think that it is somehow problematic. By definition you need to return much more than the VC gave you on the profit route. If the other route is “only” to apply for money by other non profits that would be a problem.

@stevekrouse Have you written a proposal for perspective investors (non-profit or otherwise)?

@stevekrouse have you tried GitHub sponsors? It's like Patreon but the bonus is that if you get in the beta Microsoft will match the first $5K you get: https://github.com/sponsors...The $5K match is small, but gives you incentive that could lead to something interesting....