[danpalmer]

Please don't take this as a personal attack, but I see this sort of attitude as the crux of the tech "disruption fetish". I'm guilty of the same thing often, but I have to push myself to see past the tech solution and to the human expertise it is replacing. Also this isn't so much a defence of cabbies, I don't use them and think them expensive an antiquated. It's a defence of the process being a better one than the relatively basic tech solution we have so far.

Google Maps just isn't that smart yet. Yes, you can absolutely drive a car through London on a route yourself and it'll work fine, for some definition of "fine". On the other hand a taxi driver is supposed to be able to perform to a far better definition of fine[1].

Google Maps accounts for traffic when it's bad. Does it account for traffic patterns before it gets bad? Maybe, badly. Does it account for bad traffic that it knows is going to happen because it chatted about an event in a certain place with another cabbie over tea earlier, or because it knows that there's a road closure for a race this afternoon? No. Google Maps is so far from taking into account things like this, but they are natural for a human, particularly one with expertise.

Google Maps will take you to a place, but what if the place you need is more nuanced than that? The stations in London often have multiple entrances, and walking from the wrong entrance to your platform could be 5 minutes and you'll miss your train because of it. Cabbies can take you to the correct entrance because they know this. Does Google Maps account for this? Maybe, badly. What if you need to get to the stage door of a theatre? This could be a few minutes away from the public entrance, but cabbies are supposed to know these sorts of landmarks. Does Google Maps account for this? No. What if you want to go to The Dorchester, let's hope you get directions to the hotel and not the town.

It's easy to excuse getting the wrong entrance to a location, or being off by 15 minutes on a journey, when it's just you driving yourself through London, but this doesn't work at the scale of public transport, and doesn't work if you are buying a service from a provider who is supposed to have intimate knowledge of the city.

Tech can probably manage these things at some point, but like with so much "disruption", we've solved a basic version of the problem and declared it better and stopped listening to people who have collectively been doing it for a century.

[1]: Some taxi drivers in London are crap because they can't be bothered. I'm talking about the hypothetical good, qualified, cabbie.

[jimnotgym]

Tech often does this, it takes a thing, deskills 95% of it, and puts the people who can do the missing 5% in a horrible situation where they can't make enough money to live. Instead they see other people making more than them by half-assing their profession.

I was a Carpenter and Joiner. Crappy gangs with nail guns and battery saws could compete with me on some of my work, leaving too little of the skilled stuff to make a living. CNC stair builders churned out a demonstrably worse but cheaper version of what my workshop made. My Dad, long past retirement age, is still living on the crumbs these kids leave behind. When he dies there will be none left in our old patch that can do some of the old skills, like traditional roofing (where prefab trussed rafters can't be used).

I am not 100% clear that this is a better world.

I use the tube in London, but black cabs are still the premium service.

[Kalium]

> CNC stair builders churned out a demonstrably worse but cheaper version of what my workshop made.

That sounds like is is better by the standards the customer cares about. The trick, I think, is realizing when your goals and those of your customer diverge.

[throwaway2048]

The problems with ignoring every signal but price should be obvious.

How many people understand whats important in a roof joist when they are having their new home built?

[Kalium]

I think a more subtle, and perhaps salient, question is how many people not trying to sell something understand what's important? It's very easy to mistrust the motives of someone trying to do customer education.

You're right. It's obvious that price isn't the only signal. But perhaps it's more readily comprehensible in many cases, especially when other signals are potentially less trustworthy.

[jcrawfordor]

Far from London, in San Francisco I found that yellow cab drivers could often (but not always) take me to where I wanted if I gave a cross street or described an area of the city. Uber and Lyft often couldn't because Google Maps is surprisingly unreliable when it comes to searching for cross streets. Just my house was an example, yellow cabs I could usually tell the cross street and they knew it immediately, and Uber/Lyft I usually had to give the full address and then wait while they typed it in. Part of this is because humans can retain a lot of location-specific intuition - I lived at the intersection of a regionally important through street and a university. Cab drivers knew the through street, and knew what part of it the university streets were in... the rest was just driving down it until they saw the right university.

I also had a LOT of experiences with Lyft drivers that were clearly very new to the city and lacked basic awareness of unusually set up but major intersections, and near-rush-hour traffic patterns. This lead to things like going the wrong way on one-way streets and missing turn lanes.

Software still has a long ways to go before it can adequately capture and work with that type of region-specific organization.

[jennyyang]

Sorry, I think you are making this up because it sounds like you've never actually used Uber before.

> Uber/Lyft I usually had to give the full address and then wait while they typed it in

With Uber and Lyft, YOU have to type in the address in the app before the car even gets summoned. You also get to see the destination so that there is no mistake. So I don't know why on Earth you would think you have to wait for them to type it in. You can't just hail an Uber and tell them where you want to go once you get in. That's simply not how it works.

[zacwest]

That is how it works now. That is not how it worked in the past, and presumably that is what they are referring to. I have many memories of jumping into an Uber Black in SF (before UberX existed) and saying cross-streets. It took a while into the evolution to add the destination picker, let alone make it required.

[jennyyang]

So you admit entire post is a lie, fabricated on information that is 10 years old. Gotcha.

[pbhjpbhj]

Someone must have done a cabbie vs. satnav challenge surely?

Does Google definitely not account for future events? If I plan a journey for London Marathon day, say, does it not even warn me of delays?

Does it take in warnings of engineering works (eg for rail travel)?

[notatoad]

>someone must have done a cabbie vs satnav challenge

I'd argue it's been happening on a fairly large scale for the last few years. That Uber is competitive is a good sign that the Knowledge isn't nearly as much of an advantage as the cabbies would have you believe, but nor has it been completely eclipsed by tech

[danpalmer]

I wouldn't be surprised if Google accounts for the London Marathon. Does it account for the opening night of a new show in the West End? I highly doubt it.

I don't think Google maintain their own data for rail travel, and the data sources aren't that good, so I doubt they'll be accounting for engineering works unless either they are scheduled far in advance or they've been happening for a while.

In my experience it takes time equivalent to a bunch of people having a crap journey and Google collecting that data for Google's instructions to change, they're not pre-emptive based on expert knowledge, they're reactive based on collected data.

[raspyberr]

They have. Just search on YouTube. The ones I could quickly find were about 10 years old unfortunately and not exactly extensive experiments.

[Kalium]

> stopped listening to people who have collectively been doing it for a century.

My butt has been a butt for a few decades now, and I don't have to listen to a thing it says. Perhaps mere duration does not automatically imbue a practice, habit, arrangement, or person with value and wisdom worthy of preservation.

If we came upon the idea today, and someone told me they had memorized a ton of streets, could sometimes do marginally better than GMaps, and therefore deserved to be paid a major premium over what GMaps could do, we would be skeptical of the cost/value tradeoff. Except, perhaps, as a luxury service for the upper crust.

[xadhominemx]

Waze does a much better job of accounting for traffic than cabbies. It's true that sometimes cabbies will drop you off at a better door - is that worth 30 pounds?

[jimnotgym]

I don't think that is necessarily true. I often commute 40 miles on relatively quiet roads, and then at the end of that another 10 miles on one of the busiest roads in the country. When I leave home Google and Waze both underestimate the delay caused by the fact that the busy road is gridlocked by the time I get there. I am therefore better than Google and Waze at calculating journey time.

[danpalmer]

It's not worth that to me no, but saying that Waze/Google are better than a human is a different thing and I doubt that's the case. I think like many problems, doing this well generalises to Artificial General Intelligence.

[commonturtle]

These are little optimizations though. I see asking taxi drivers to pass The Knowledge as the same as asking students to use logarithm tables or memorize digits of PI: yes there is some marginal benefit to doing it, but honestly there are so much more valuable things one can do with their time and energy.