Hi all! I'm When. I'm excited to participate in discussion here and share ideas. I've been a long time dreamer in what has been possible for the future of code. I've been working in software for over 20 years now and I've always been passionate about empowering people with software, especially those who don't know how to code. I have a degree in both Art and Computer Science, and the intersection between expression, intention, interface, abstraction, and usability has always been of keen interest to me. I mostly found my calling in writing tools that allow other to work at a higher level. In 2007, I wrote what I believe to be the first ever "reactive javascript framework", at least in the form you see now - client-side templates, data-binding, client-side routing, components etc. I had to do a lot from scratch: module system, template compiler, fine-grained data-binding system. I've been watching the rest of the industry do it wrong ever since (I'm kidding! 😝). It's still in use by the company I built it for (although I don't work there anymore), and it beat out Vue 3 + Vite in a head to head evaluation recently (from what I hear). It currently powers a loan automation system that handles billions of dollars in loans. I was never able to open source it, so I guess y'all have to take my word for it, lol. Happy to talk about it if there's anyone curious. It still has a few tricks I haven't seen in the wild. I've also built a custom workflow engine and business rules system that had a pretty unique architecture and scaled really well (over 500k rules), and actually delivered on the promise of moving a huge amount of logic into the hands of non-programmers (without needing programmers to babysit it). Those are some things I'm most proud of and have deeply informed some of the ideas that I imagine for the "future of coding". My closest thing to an actual claim to fame was my involvement in the standardization of ES6. I have a name credit associated with breaking the standoff around classes. The "maximally minimal class" proposal was based off of a proposal I made. These days I've been working on my own vision for the "future of coding". I'm a big fan of Alan Kay and Brett Victor. From Alan Kay, I really love the idea that we need to shift from mechanical to ecological thinking. From Brett Victor I love the concept that the "digital medium" needs a separate concept of "authors/authoring" that is separate from engineering/programming. The fact that programming languages have been a tool for both is part of what holds us back. I'm working on a unique sort of universal IR binary format inspired by DNA, and a programming/modeling language inspired by cellular biology. Happy to chat about any of this stuff, but mostly here to see what everyone else is doing! Looking forward to learning from all the brilliant minds here.

I'm in awe! We had a thread about workflow engines a few weeks ago, which feels like a blindspot to me, would be curious to read more from from you about it

Yeah, sure, let me see what I can do to convey this stuff. So first of all, the system I built was not based on BPMN at all. It was built from scratch and adapted over many years to suit the wide variety of use cases needed. It was still a reasonably generic workflow system, but by being something in house we were able to make it have the actual features that would get the job done, as opposed to grabbing something off the shelf and praying. My core criticism of BPMN and most workflow systems - like say Camunda - is that they are simply a kind of glorified chart execution system. In general, they are completely disconnected from the actual system they are meant to control. They started from the same kind of modeling background as UML, and have historically had trouble IMO by effectively standardizing first. There was a lot of speculation about the value it could bring. On paper it might seem like the right path: create a visually oriented tool for describing business processes. Focus on the visual language that subject matter experts might be able to understand. Then create a standard for it so that a variety of tools can consume it. One of the problems you see is that the early versions of UML/BPMN were really about a shared visual language for communication purposes. Then as the hope of making it work through automation came into play and a variety of enterprise vendors got involved. Both UML and BPMN went through revisions into order to give them execution semantics to actually run. It was also happening during peak XML/WebServices complexity explosion was happening. This also dramatically increased the complexity, in many ways making a kind of worst of both worlds. Its no surprise to me that the AWS step function system is not based on BPMN, but is quite simple in comparison. If you want a gruesome inside look of what creating a standard through OMG is like, I find this tale about CORBA to be a good example of the kind of insanity that happens when you make a standard from vendor soup: https://queue.acm.org/detail.cfm?id=1142044 there's a lot of "castles in the sky" kind of thinking. Here's an account around UML that echoes the similarities https://tratt.net/laurie/blog/2022/uml_my_part_in_its_downfall.html. A lot of dreaming of how to model the happy path of processes without actually building something and using it and making sure it actually worked. A lot of kitchen sink acceptance of feature requests without a simple coherent vision. When I built our workflow system, it was a completely different sort of process. We started with the simplest thing we could build - a kind of simple flow chart/state machine that was built as a functional system first. As we built out the larger system (like I said, it was a complete loan automation system), and as we were trying to handle the problems of a real national mortgage lender, the system was built to solve the actual problems in ways that would actually be used/useful. Where it ended up was dramatically different than something like BPMN.

Thank you for going into great detail, I really enjoyed that (and all the links you shared)! The kind of perspective i was hoping to get here. Also about UML which I've vaguely associated with all this, but it was fascinating to read about the standardization efforts

yeah, I think its kind of interesting how IMO some really fumbled attempts at a kind of "model-driven development" have completely soured the entire concept for many people. I think there's a lot of potential value in it being done in a much better way

ha fun connection that I can relate to, I feel very similar about the RDF stuff. I have a livelong affliction of being too castles-in-the-sky so all my side projects have been shots at immediacy. For me Bluesky and its ATProto are great modern examples of rubber-roading first and then standardising it (tho I guess that success is still TBD)

yeah, I agree about Bluesky. Yet to be seen, but if its gonna happen, I think thats exactly the sort of path that will do it. I'm honestly a lot more skeptical about the long term success of ActivityPub in comparison. It's not as bad as RDF, but I think there's some inherent flaws that will prevent it from really succeeding at scale. I just don't think that federation of state in that way is the right approach. I guess we'll see! At least it has a committed working implementation, and started from implementing before standardizing.

Yeah I share that worry, ActivityPub seems designed for more nimble use cases which is cool for all who need that but was not what I was looking for when I was shopping for a Twitter alternative

So I'm actually really interested in figuring out better decentralized system in a broader sort of way - like how can we make decentralized more the default. I think its interesting how some of the early choices of the web really lead us here. For example, making it so that the http protocol was stateless and simple was an understandable choice for a kind of thin client, public document protocol, but it also creates the kind of systemic forces that force every server to own your data and identity and trends towards centralization and winner-takes-all dynamics.

I share that interest! If you haven't seen it already, I quite enjoyed the discussion between a Bluesky engineer and a Mastodon co-founder here (that's the last response in the thread I think, but the previous ones are linked). Christine who authored that one is afaik also working on something that attempts to answer the big issues you are raising (I haven't read about it yet tho, so not sure how successful, if I can even judge that)