Hacker News Reader

Comment by rdl | Hacker News Reader

We have $2/gallon gas, from massive domestic production (and similar or lower prices for Jet-A). There are traffic congestion reasons for mass transit in cities, but we're pretty sorted for inter-city.

View on HN

The article talks about a train that can go 200mph. That makes having people drive between cities seem a rather slow option. Plus you can get work done on a train. On a car it is not possible as a driver (other than phone calls) and pretty hard as a passenger.

I take your point about planes, but it's good to have more than one option and trains are good for middling distances, as there isn't all the need to check in, and they can start and stop closer to where you want to depart from and arrive.

Not to mention global warming :-)

Why just 200mph

Because it's here today. It's easy with simple investment, it's got huge benefits over the joke that is the current U.S train service.

There are plenty of passenger trains in service around the world that go faster than 200mp/h, today.

Those are all in China. Wikipedia tells me that max achievable speed in regular service in every country except China is 320km/h. China does 350km/h, but you need to be traveling at quite a long distance for a measly 30km/h increase in speed to be worth it, and China runs much longer high speed rail services than other countries (Beijing to Hong Kong is 9 hours).

Besides, the more important thing is average speed. Acela's top speed is respectable for legacy rail, but it achieves this top speed for a total of 33 miles over a 457 mile route.

> Those are all in China

You're probably right. America shouldn't aim to achieve speeds of high speed rail in ~2025 that China achieved in ~2015. That's aiming too high.

Best to aim lower so we don't disappoint anyone.

And China pretty much stopped running train at 350km/h.

Steel-on-steel rail is realistically limited to top service speed of 320km/h, maybe 360 at most. The maintenance is just too high when you increase speed beyond that. Newer Japanese Shinkansen line only operate at 260km/h, for example.

The most painful part of high-speed-rail, particularly at higher speeds, is the land acquisition costs; minimum curve radius grows exponentially as speed rises, so costs escalate as speed gets higher. China cares far less about compensating property owners for fair value, which is why they can engineer such high speed lines in the first place.

We are talking about the difference between 217MPH and 200MPH. The total travel time saved between Boston and Washington by an additional 17MPH is a whopping six minutes with the generous assumption that you start right off the bat at top speed and maintain it the entire trip. If saving six minutes will cost billions or tens of billions of dollars, why bother with such a marginal improvement, when you could spend those billions on literally anything else from housing to healthcare to upgrading the power grid?

Yes, but the US just needs a taste to get going. They don't need to aim too high, they should concentrate on delivering something.

As you increase speed, you need to take turns at larger and larger curves to reduce g-forces on passengers, and the curve radius increases exponentially vs speed. The bigger the curve, the more land you need to buy, and depending on which country you're in property rights extend above and below the land so the cost is unavoidable. (If you can avoid it by going underground or overground, the cost of tunneling and bridges is expensive.) It becomes increasingly costly to increase speed, and on top of that you get diminishing returns with travel time (speeding up from 80 to 120MPH is worth a lot more than going from 200MPH to 240MPH). And above a certain point you spend too much time accelerating and decelerating from top speed, and not enough time actually at the top speed to make it all worthwhile.

Regardless of motive technology, curve-radius will always be a relevant problem, because one of the major advantages of the train is that you can walk around and about. We could design trains with sharper corners and harder acceleration/deceleration, but no one wants to be strapped into a vomit comet for three hours.

> and the curve radius increases exponentially vs speed.

It's merely quadratic if I recall correctly

Oops, I misspoke. But yeah, it's a non-linear increase in costs that results in increasingly marginal time savings.

Instead of trains we're working on self driving cars. They better serve last mile and are better suited for private investment and innovation than huge infrastructure projects.

Can you work in them without getting sick?

The self driving car can't go 200MPH.

You also can't fit as many people in self driving cars, simply because having the requisite fuel storage and engine per 2-4 seater car is quite a lot of space when you can spread that around 200 people per train car. And this is before we consider that average car occupancy is something like 1.5.

When driving your car inter-city do you never think ‘if only I could be working, relaxing or sleeping with this time’ like in a train?

Not really because it takes me a fraction of the time. I can do those things far more comfortably when I get there.

Self driving will solve that; self driving is absolutely there already technically if we restricted it to special lanes/highways/etc, and probably could be 100% of interstates already too.

There are some city center to nearby city center where rail would be nice, I agree.

Most of the time when I'm going intercity it is by plane.

If you think you see general solution for "self driving" in next 20 years, i have a bridge to sell you

In a restricted lane (dedicated to only self driving vehicles), we have had this capability for at least 15 years.

I hope no one is working or sleeping in those lanes.

That might be fair but that's a chicken-egg problem.

We don't have any such lanes because only the self-driving car companies have dual mode vehicles.

Nobody's going to market vehicles designed to drive in lanes that don't exist. And building out the lanes and the infrastructure to enter and exit them, probably will boost congestion in the short term.

Come on, the train is already semi-self-driving.

Self driving car will never match the safety, speed, comfort, and social interactions of a train.

My commute is under 15 minutes each way and this is the only thing I think about during that time. I hate every minute of attention sucking driving.

Part of my day is a 20 minute train ride. I tend to enjoy it (except, of course, when it's late).

Traffic congestion is a pretty big issue for intercity too in some regions. Dallas-Houston is a mere 240 miles but often takes 5 hours to drive. Even in "light" traffic, 4 or so is common.

You can do a lot better than that with even decently upgraded "legacy" rail, never mind a Shinkansen/TGV type system. For example, London-Manchester is 210 miles, and trains take only 2 hours.

Is there a solution to the last mile problem? It's every difficult to get to the final destination in sprawled suburban cities even if there's a direct center-center high speed rail line.

Manchester has a great central tram network. London has the underground.

Driving in London is mainly an exercise in stress and pointlessness.

You hop in an Uber.

Build dense cities.

For example the Tokio-Osaka Shinkansen averages 130 mph over the whole route with 15 stops in between the end stations. You could get from midtown Manhattan to downtown Boston in 1h40min. Make that 1h15min, if the train didn't have stops between NYC and Boston.

So visiting Boston could be almost as fast as visiting the far end of Queens, from Manhattan.

> London-Manchester is 210 miles, and trains take only 2 hours

Yes and to further emphasise, this is not a high speed train (though it does use leaning trains making it faster than other lines, and has few stops. Also it's 2 hours 15 mins to be precise).

The proposed high speed train promises London-Manchester in ~1 hour.

It's only 2hrs 15mins if you go the longer way via Crewe, the more direct services via Stoke with fewer stops are around 2hrs 5mins. It also depends on your definition of high speed train, the original definition is 125mph, which the Pendolino definitely is (and is actually the max speed of the British train called the High Speed Train). We keep redefining what High means... HS2 with a max speed of 186mph will be 67minutes.

Good points. By the way just for interest - I take this service regularly and for me it seems to take 2:15, going via Stoke. Furthermore I've actually never seen it go via Crewe, I didn't even know it did. This might depend on the time of day though - I always take it at the busiest times and/or weekends which might increase the journey time a bit.

The danger when drawing comparisons between England and the US is scale. England is slightly smaller than Alabama.

Overall yeah, but that's why I focused on similar-scale regions. For long distances, like SF-NYC, I think the current US solution of planes is basically the practical one and am not expecting a coast-to-coast bullet train. What's slow and could be improved in the US is regional-scale travel, the 150-400 mile range. I was disagreeing with the person who started this thread who argued that that kind of intercity travel in the US is already solved by autos. It is sort of, but only if you don't mind slow and tedious travel...

For inter-city travel high speed trains will outpace any car even under ideal traffic conditions.

Paris - Stuttgart Car: >6h; TGV: ~3h

What about climate change?

Plant a lot of trees using the rail money?

Since cars have a 10-20 year lifespan, and electric is becoming superior, that will probably solve itself over the next 10-20 years (maybe they'll be EV powered by natural gas for a while, but that's still a win, and then the grid might get greener.)

Aviation fuel is the highest and best use of hydrocarbons (along with some chemical production), so we can deal with that last, after lots of power generation, demand reduction/efficiency, etc

Airliners have gotten far more efficient over the last 40 years, starting with the 757/767. (The 757 design got its start from more efficient engines and a new wing shape.) In fact, efficiency improvements are the primary factor driving new airliner models and sales. At the time of the 757 launch, fuel was 40% of the operating cost of air travel.

A huge improvement (for the biggest planes) is going to twins and ETOPS (not just fuel, also maintenance cost, but also fuel). Not as relevant to domestic routes, though.

Not sure your math checks out. EVs are < 1% of fleet and < 2% of sales in the USA. As you say, cars last 10-20 years and this number is rapidly getting longer. Vehicles older than 16 years are the most-rapidly-growing segment of the US vehicle fleet. The average age of US light passenger vehicles hit 12 years in 2016. In 2020 there will still be 80 million cars on the road that were built 1995 or earlier, none of which were electric or even hybrids. Virtually all of the cars sold this year will still be on the road in 2030, and 98% of them will be combustion powered.

What matters is miles driven, not number of vehicles. For someone with a 1-2h each way commute (sigh), replacing a 2010 truck with a 5-year-old 2025 EV in 2030 will make sense on a 2 year cash return basis. For someone who drives 1-2k miles/year, it probably isn't worth upgrading.

Yeah but EVs make up a far smaller proportion of miles driven than they do the fleet by units. For obvious reasons: many EVs are short-range vehicles and the people who buy them are self-selecting as people who are both environmentally conscious and able to live with the range problems. People with long commutes who can't afford a Tesla are obligated to burn gas.

It is reasonable to assume that a 5 year old EV will be as cheap in 2030 (2025 model) as a 5 year old EV is today (2014 model), while becoming more capable at roughly the rate we've seen so far. A 5 year old Leaf is $6-10k now, and would be a ~250 mile range car by 2025.

If you have, have access to, are able, or want to drive a car, then at best it's reasonable option for places outside of denser areas, albeit far more expensive than the cost of gas. While cars are certainly good in some ways at getting people around to places that don't make sense to service with more significant options, they shouldn't be considered first when thinking about new development imo.

A non-new, efficient car (say, a 3 year old Prius) is about $0.15/mi to operate.

Perhaps, though I don't know where that number comes from. If it's purely gas mileage, then it doesn't account for cost of insurance, cost of the car, risk of an accident, cost of storing the car, cost of general maintenance, and, while sort of ancillary, cost the urban area if everyone relies on driving. If it's not pure gas mileage, I'd be curious how that formula works out, but there remain problems with that as a solution to anything.

That's all-in cost, including depreciation and fuel and insurance and maintenance. This is why UberX sort of works for some drivers.

The rate for a new car is much closer to the federal reimbursement rate ($0.545/mi), mostly depreciation. Obviously driving a new lambo at 100mph will be more like $4/mi.

I'm really not finding how this can work out to $0.15. Could you clarify about what the federal reimbursement rate is? Again, if that number were true—which it may be but so far it seems unlikely—the remaining issues are important. AAA, CAA, and other news sources, it takes a very generous amount of miles to get there.

How is that number calculated?

AAA and others have good data (usually for new cars, though, so depreciation makes it $0.30-0.40/mi total).

https://therideshareguy.com/is-a-toyota-prius-the-best-car-f... has a 2011 and 2016 Prius.

I haven't gotten a chance to read through that yet, but even if a Prius is the best ridesharing car, i doubt the best car for an Uber driver and the best car for someone driving 10 miles to and from work five days a week are going to be the same. You'd have to drive a real lot to have the increase in gas mileage cancel out the higher purchase price and higher insurance costs vs a basic econocar.

> at best it's reasonable option for places outside of denser areas

AKA 99% of North America.

Places yes, not population though.

Thanks to subsidized exploration, and a federal highway tax that is 30 years behind, and no accounting for negative externalities.

Reply on news.ycombinator.com