4 ms·
It's definitely a great example of an engineering/science project run by grownups. Compared to what I see in the daily cloud software engineering world...
by dekhn 5y ago
It's definitely a great example of an engineering/science project run by grownups. Compared to what I see in the daily cloud software engineering world...
- bdamm 5y agoWell, if money is no object...
- lmilcin 5y agoAnd if moving schedule by a decade is not an object...
- boringg 5y ago
- jve 5y agoYour average budget and timeline in daily cloud software is certainly multiple orders of magnitude less.
- stackedinserter 5y agoDifferent requirements, different engineering. Nobody needs space-mission-ready CRUD that takes $10 billion and 15 years of development.
- JaimeThompson 5y agoI wonder how much has been spent on electronic medical records and related software in the United States?
- ajmurmann 5y agoThis is an interesting example as the domain is probably closer (albeit still far away) to the JWST than a photo sharing app is. There is a very high minimum bar to get anything into production because of high regulatory requirements. I bet iterating in this area is much slower and harder because of regulatory hurdles. Definitely an area that requires different processes than your typical SaaS web startup.
- dekhn 5y agoThe number is incalcuable, see https://news.ycombinator.com/item?id=13861074 https://news.ycombinator.com/item?id=13861074 for some context on the fractal horror that is Epic EMR.
- deniska 5y agoIf spinning up a new microservice would be as complicated and expensive, as yeeting fragile scientific equipment to a Lagrange point, you'd see better engineering in the cloud software world.
- dekhn 5y agopittsburgh?
- ajmurmann 5y agoIt's because the constraints are so different. Deploying to prod is incredibly expensive in this case and you practically cannot iterate outside prod either. I see this every day in the difference between SaaS and on-prem Enterprise software. If you have a bug in on-prem, Enterprise every customer needs to deploy the patch. You need to update all affected, supported versions, etc. In SaaS, you just deploy and more often than not that's it. This changes the way you do everything and it comes with a high cost. JWST is the most extreme example on this spectrum.
- dekhn 5y agoI helped run a world-class MySQL installation for one of the world's largest tech firms. The install was the most mission critical system at the entire company and downtime was extremely expensive. I was hired because during failovers, the engineers were often getting unhandled runtime exceptions in the failover python code. (much of it was untested except when running failover) Often, the engineers would break one thing while cleaning up another thing. All of this was OK, because the system failed infrequently enough and was eventually replaced with spanner.
- krisoft 5y agoI wanted to say that for 10 billion USD I could also deliver you miracles in the "cloud software engineering world", but the truth is probably not. If a pure software project would get that kind of money people probably would feature bloat to infinity. The JWST during its design phase were under very strict limiting factors. Volume limit, mass limit, power limit, etc. Compared that with any software project where obviously there are limits but they are much more malleable. You can argue that we need to provision 20% more servers to achieve our goal, but you cannot argue that we need to stretch the Ariane 5 by 20%. (Well you can argue, but the rocket is not going to get stretched.) So all in all, it seems I'm agreeing with you. But I don't think it's about the quality or grownupness of the participants, but more about the shape of the problem.
- dekhn 5y agoThe way I think about it is the standard was set by the Apollo engineers. I think many of us are familiar with the MCP and all the other cool details about the hardware of the mission, but the reality is: everybody who worked on that had previously done something similar, but for the military, and they took every bit of hard-won experience from the previous projects (ballistic missles guidance, military radar, data coding, etc) and applied it in a totally unceremonious and highly effective way. I think the modern computing world instead is not grownup, but rather CADT (cascade attention deficit teenagers). A constant developing of new solutions that copy previous ones, but are typically worse, or have the same problems as the previous ones. This seems endemic in the industry, albiet with exceptions.
- jacquesm 5y agoQuite. It will be a long time before the software industry will grow up at the present rate.
- dekhn 5y agoThere are plenty of subsets of the software industry that are grown-up. Anybody who wants to learn more should read the RISKS mailing list and its archives. It's remarkably hard to produce reliable software but a number of groups around the world can do it. Examples I can think of: the teams that built the series of ESS for AT&T. The folks who built Erlang (telecom reliability is a common theme here, being one of the other technical areas where software and hardware engineering excellence greatly exceeds that which we see from typical groups today). Even Qt, although the quality of their code and its adherence to the principles that make Qt great is variable, I'd say is a good example of a system built by grown-ups. BTW, when i say "grown-up", I don't really mean anything directly related to age, although age and experience are the best proxies for grown-up. Grown up is really a way of saying: working with more weight on wisdom as raw intelligence (Ghemawat's law), respecting experience even if it's not obvious why you should, (Chesterton's Fence), using statistical testing techniques effectively for quality control (Shewhart), practicing discipline to ensure long-term compatibility and not burning your dependents (jwz's law of CADT). I don't think there's a general term for acting with these principles, I just mentally associate them with being a grownup who has learned from experience.
- voz_ 5y agoI cannot stand this kind of comment. It's worse than nerd sniping. Rude for no reason, not curious, not intellectually stimulating. Reddit level drivel. Perhaps a better question would be - for public or enterprise facing projects of similar budget, why is the perceived $X/$Y/$Z so much lower? That is a discussion worth having, and can help unpack your bias and assumptions into a worthwhile conversation.
- dekhn 5y agoAs other people have replied, differing missions and costs of failure explain most of this, although in my opinion CADT in web development is actually the primary problem. I won't link the jwz's essay here for good reason. In short, the expansion of the web economy has allowed for incredibly lazy programming, being done by minimally qualified people, mostly motivated by short term gains, who consistently forget the expensive lessons of previous generations. The original web technologies were fairly good- HTTP and HTML both addressed a need, and I think they were sound (never liked html's derivation from sgml, though, and I don't agree with HTTP's decision to be stateless) but so much of what's sprung up since has evolved with no real natural selection.
- dang 5y ago"Eschew flamebait. Avoid unrelated controversies and generic tangents." "Please don't post shallow dismissals, especially of other people's work. A good critical comment teaches us something." https://news.ycombinator.com/newsguidelines.html https://news.ycombinator.com/newsguidelines.html
- dekhn 5y agoNone of this is flamebait. It's completely correct to say that modern software development has evolved into clown cars and that big infrastructure projects are run to far higher standards. Also, why can't I downvote your comment?
- grzm 5y ago> Also, why can't I downvote your comment? You can't downvote a reply to your own comment.
- dekhn 5y agoHuh. it won't let me downvote anything in the subthread. Just upvote. Ugh.
- dang 5y agoIt was a grandiose, information-free putdown. Those are flamebait. It also included name-calling ("grownups") which obviously multiplies the flamebait. The issue isn't how correct you are or aren't. The issue is comment quality. Your comment was low-quality regardless of how correct you are, or feel.
- dekhn 5y agomeh.