So, are you arguing waterfall is the better process for innovative designs, and agile should be used only for small renovations?

I think I agree with that a bit. For hardware, Apple might be an example showing that dedicating significant amounts of research to your designs is more likely to lead to good hardware designs.

I don’t see that happen soon in the software industry, though. We’re still constantly replacing libraries and frameworks with stuff that promises to, eventually, be perfect because the current approach isn’t perfect.

This seems a bit defeatist. Some hope for and striving for improvement is good for a society.

There is improve and there is trying something new. Rails have been around for nearly two centuries - they are a solve problem. We don't need to innovate on many parts as there is nothing wrong with the old technology with some minor updated. We already know what works well for a standard train system. Innovation generally tries things that look good to politicians who don't understand trains, but in reality people who understood trains 100 years ago had the idea and rejected it as a bad idea - and the reasons they are bad ideas didn't change!

there is a lot of room to improve trains. However it all starts with standard gauge steel track, with standard sizes tunnels, standard switches, standard control algorithms (with computers this had recently had a revolution, while there is room to improve this is again in the mostly solve area), standard "rolling stock" and so on. Let me bring out here two standards that are a bit new: platform doors - now that someone invested them and proved they work everyone should retrofit to them; and elevators - it is criminal that many stations around the world still haven't been retrofitted (criminal as in I think someone should be in prison anywhere you find they have not been retrofitted) Once you have all the standard stuff laid out you can make a few tweaks to see what works, but only a few as we know the standard stuff works.

Make sure when you look at standards above you look at what others have done. If you have an idea odds are someone else has already tried it: find them and figure out how it works - sometimes you can tweak to be better sometimes you realize it is a bad idea.

Rails in particular have improved quite a bit. Modern newest generation slab designs requiring very, very little maintenance and are fast to install.

> Innovation generally tries things that look good to politicians who don't understand trains, but in reality people who understood trains 100 years ago had the idea and rejected it as a bad idea - and the reasons they are bad ideas didn't change!

I'm all for respecting old ideas but this idea that modern engineers are idiots and can't think of anything people 100 years didn't think of is also wrong.

> Make sure when you look at standards above you look at what others have done. If you have an idea odds are someone else has already tried it: find them and figure out how it works - sometimes you can tweak to be better sometimes you realize it is a bad idea.

I don't disagree.

In the context of Crossrail, this seems to be what they have done for the most part. So I don't understand the criticism.

> is idea that modern engineers are idiots and can't think of anything people 100 years

That is not what I was trying to say. I'm saying that politicians and managers are unable to think of things that people 100 years didn't think of. While that isn't 100% true, it is very close.

Though most of the time engineers today couldn't think of anything they didn't think of 100 years ago. However sometimes advances in other science means ideas that wouldn't work back then could work today. This is almost entirely incremental improvements though.

Crossrail didn't follow the same practices as building 100 years ago. And the results are terrible. (of course some of this is good - better safety standards - but the whole is a system that costs far more than expected for lower quality than possible elsewhere using modern standards)

Crossrail uses advanced signaling making very good use of the resource. They use platform screen doors for both improve safety and reliability but also fast loading and unloading. Seem to me these are all improvements and were the right approach for Crossrail. They didn't do everything perfectly but the system is pretty damn good and pretty damn successful.

Yes, infrastructure has gotten more expensive, and that is particularly true in the Anglosphere. But those problem don't have so much to do with the chosen design, but rather it operation and planning processes.

What actual parts of the design are actually bad?

I'm not up on all the issues with Crossrail, but that trains cost too much in the English speaking world is well known and Crossrail is not exempt. Generally this comes down to stations that are much larger than needed and other things that appear to be justified until you dig deeper and discover they are expensive and the rest of the world does without just fine.

Trains cost to much because like many other countries, Britain buys from their own manufactures. Because of the state of British railroading since privatization most of British rail manufacturing collapsed and only a very small number of manufactures are left.

Station being to large is a question of taste and priorities. Saying that such tradeoffs are not worth it is different then just saying straight up that 'the results are terrible'.

I feel the point is a bit subtler. We should strive for improvement. We should also understand that that takes more than the normal amount of resources to do something different and better & either commit to spending what's required or, if we can't afford to, use something already proven instead.

The half-done or broken improvements still get called "innovations" and give innovation overall a worse reputation.

> I feel the point is a bit subtler. We should strive for improvement. We should also understand that that takes more than the normal amount of resources to do something different and better & either commit to spending what's required or, if we can't afford to, use something already proven instead.

Indeed. I wish this point were articulated that way more often.

> The half-done or broken improvements still get called "innovations" and give innovation overall a worse reputation.

We fetishise innovation as an abstract concept, for good reasons because you cannot have progress without innovation. What we tend to miss is that while innovation is good, some specific innovations are terrible. This is particularly infuriating when they are re-surfacing old, solved problems. Looking at wall tiling for example (or the cladding that can be found in much of the Underground): this was a good solution to a common problem. Bare concrete walls have obvious downsides, which is why we don’t tend to use them in building anymore. Doing away with tiling or cladding sounds like an architect being innovative, but if the replacement does not solve the problem, it’s a regression. It will look modern and clean as long as it will be properly maintained, which is to say for about a year, and then it will decay the way concrete does. It will eventually be clad, or reviled as a post-modern monstrosity the way some bad brutalist buildings are today.

Innovation for innovation’s sake is cargo culting progress. This is giving innovation a bad name.

What innovation is left? Concrete has been around for centuries in modern forms - we can tweak the formula a lot, but still the same basic idea as when it was invented 100 years ago. Most of what passes for innovation has been tried before. Other things were not tried because anyone who understood the problem knows they are cannot be cost effective. Unfortunately the people who fund this cannot be experts in the field, and thus can easily be persuaded to fund things by scammers that anyone who really was an expert could tell you in a few hours of study are a bad idea.

The problem isn't that the politicians are not experts in construction - there is far too many things politicians fund for them to become experts in even a fraction of them all (they also have to fund medical studies, military spending...). The problems is politicians don't respect trained experts, thus don't keep people around to become deep experts, and in turn there is nobody around with deep technical knowledge to say what is a good idea. Or if there are such people around they are not in a position where they can be listened too, instead any scammer is allowed to present their pitch for "innovations".

Or you know, be innovative learn what works and what doesn't. With each project improving the state of the art forward.

The real cost of these project is the big tunneling, the trains and so on.

Going for something innovative makes sense, because it can be easily changed later if it really fails (and that has yet to proven).