It's important in conversations like these to recognize that preconceptions of some whatever task are *usually* wrong. So rather than argue "2 kids in a dorm could do it in a weekend" versus "it requires Oracle and $10M", or anywhere in between, it's best to think with a little humility. My own personal approach is scient-ish. The first task isn't even to state the problem. The first task is to study the domain. Lay out the preeminent usage patterns, *then* discuss which tools in our kit might apply to which usage pattern.
If 1 use case is "sample rocks on Mars" and another is "design a UI your grandma can use" [⛧], then the development method(s) will differ. The main problem is that tech (and government, and pretty much everything else) is dominated by posturing know-it-alls who can't admit that they don't know much at all. With that, I'll admit that I have zero idea how difficult such a site will be. Were it Trump who signaled this plan, I'd believe it was pulled out of Kushner's hat with no justification. But my guess is Biden's got some team who came up with it and, if I cared, the first thing I'd do is check their creds and read the plan. There's a lot of material to use for such: https://github.com/uswds/uswds Final thought: While Biden clearly *used* to be solidly neoliberal, implying an inference like Daves, that he'd press for a government contract to someone like Oracle or whatever, he seems to be leaning a bit toward having the government work directly on big (public benefit) efforts. So I wouldn't put it past him to have the USDS do it all this time. [⛧] I had to click "View More" ... like a thousand times on Yeo and Lin's site to get to Washington. Pfft. Yeah, yeah, I didn't follow their preconceived usage pattern. So sue me. On 3/15/21 7:38 AM, Prof David West wrote: > I started as a programmer (COBOL for business apps / Assembler for internals) > in 1968. I have held almost every title up to and including CIO. I was a > professor of, essentially, management information systems but also taught the > non-math CS stuff, like O/S, programming, database, and AI. > > Spent the last part of my career making wild and fantastical claims about how > information systems could be orders of magnitude simpler and therefore > easier, faster, and cheaper to develop, with no loss of quality. The jeers > were loud and rude. > > Only had two opportunities to prove, in the real world, I was "right." One a > national company that produced forms (loan applications, disclosure forms, > mortgage documents, etc) for banks in all fifty states. They employed 50 > attorneys to review changes in law and generate requirements for programmers, > and fifty programmers using C++. My system, written in the then still quite > primitive Smalltalk) reduced the programming staff to five. The other was an > insurance company rewriting and updating a legacy system supporting sales and > management of commercial insurance (car fleets, boats, commerical real > estate). They planned a 500 million, five-year, 1000 developer effort with > multiple subcontractors and off-shore teams. Ended up being an 18 month > project with a third of the development staff (still some off shore, and one > subcontractor) at a cost of 20+ million. > > In both efforts the company had to fight a continual rear guard effort by > traditionalists. In the case of the banker forms company that effort was lost > and last I heard they were up to 100 Java programmers. > > Not a brag — I have no idea if my approach could be promulgated across the > industry with similar effect. Certainly am not claiming some kind of > philosopher stone for simplicity and low cost. > > Just anecdotal support for Jon. > > davew > > > On Sun, Mar 14, 2021, at 11:41 PM, jon zingale wrote: >> Ha, *bringing some more reality* is what I listen for to know when I have >> some naysayers on the ropes. Much of the last decade of my career has been >> working to reconcile data whose interfaces radically vary. Claiming it to be >> an 11 billion dollar problem is a rhetorical move that smells of >> *abstraction* rather than *reality* to me. Barry states the problem clearly, >> but it isn't IMO an insurmountable problem, just an intentional one with >> lots of particularities. At least one small start-up that I wrote for >> managed a similar problem surprisingly well (up to Google's data standards, >> for instance), and sold for orders of magnitudes less than the number above. >> Stating the problem is great, working the problem is best, but the rest is >> simply whining. -- ↙↙↙ uǝlƃ - .... . -..-. . -. -.. -..-. .. ... -..-. .... . .-. . FRIAM Applied Complexity Group listserv Zoom Fridays 9:30a-12p Mtn GMT-6 bit.ly/virtualfriam un/subscribe http://redfish.com/mailman/listinfo/friam_redfish.com FRIAM-COMIC http://friam-comic.blogspot.com/ archives: http://friam.471366.n2.nabble.com/
