drdunce

Applying for CS jobs is difficult full stop, with or without the degree.

I can't remember the last time I got invited to even the first stage! University doesn't fix that.

I wouldn't say it's a policy choice, the opposite. The rail system is privatised in the UK and yet still receives government subsidies. In addition, there's a huge drive to stop cars in our cities. Many "clean air initiatives", inflated parking charges, reducing lanes, pedestrianising spaces etc. It's just, regardless of how many paying passengers, or how short the journey, no one has ever been able to make trains work cost effectively here.

In the UK, bicycles are for the next level of wealth up. When you can afford to live cycling distance from your office in the city.

Same in the UK, even highly populated areas (e.g. surrounding London). Trains are very expensive compared to cars. There are discount schemes for some groups (e.g. student/pensioner), but it's really for the wealthy only.

Public transport just doesn't scale, neither does cycling. - You can't have a bus/train stop everywhere, being serviced multiple times a day when many of those locations would be visited by one or two people. - Bicycles are a fun toy, not serious transport. They are very slow and have a severely limited range for even the fittest of individuals

Honestly, not sure. But I believe hybrids only represented around 3% of cars in 2013? So unlikely to be a big effect. Especially when you consider how much bigger cars are now, and how much more they weigh.

Your new car likely weighs much more? The modern market demands a driving experience that's like sitting in our home - complete with huge sofa like seats, air conditioning, touch screen TVs, cameras and thousands of sensors. They can either cater to that or go out of business.

Carting around two tons of unnecessary random tech toys and the kitchen sink is inefficient regardless of the energy source. That's a separate problem.

Average fuel economy for cars has improved by about 30% since 2005. I wouldn't call that small.

I'd add to that: - Sensitivity to environmental conditions (current batteries only really work well at around room temperature, which isn't realistic real world outside) - Energy density (huge weight of the thing vs. the energy it has)

There's been significant improvements in ICE efficiency in recent decades. Some of the biggest leaps include turbo charging, variable valve control and electronic fuel injection and there's still a lot more scope.

Can I ask which MMO?

I don't think it's quite 200k and I wouldn't describe getting into DeepMind as easy

Again, in US? I'm from a Russell Group uni, so are friends. Most I've heard of is 37 from a friend that spent over 3 months applying and did some hard negotiating. Can you give some example companies hiring with this salary in UK? I'd be genuinely interested, as I believe would many friends/colleagues.

Maybe in US? PhD AI/ML hire in UK is ~£33k salary.

PhD in AI/ML is no longer a good bet, hasn't been for about 5 years.

I really like the various projects at https://suckless.org/

I wish there was a job board that showed only jobs worthy of giving a shit

I think the journey to the destination probably depends a lot on where you're starting out from.

For an existing/established large tech company - I'd imagine STEM PhD + MBA, 10 years experience in a senior management role.

how do test drives work?

I think the "punish all car drivers" thing will ultimately catch on. It seems to be becomming the popular opinion.

I think that's sad and damaging though. I suspect the people that are stacking up the big miles (using lots of gas) are people that can't afford to live in town and are trying to work their way up: trades/self-employed, delivery drivers, hire car drivers i.e. your junior nurse, handyman, school teacher or uber driver.

And most of them will never even be involved in a serious accident.

I disagree, it's sadly easy. You can be involved in a serious accident while stationary (broken down on the side of the road). This is an extreme example of how a lot of accidents occur when mixing speeds (being rear ended at lights etc.). It's that mix that road calming measure often create.

A plastic knife is barely a knife at all, it's almost completely useless and so is a 10mph road.

The solution to road safety can't be to drive at jogging pace. Roads are still a transport infrastructure and for practical reasons, speed has to remain an evaluation metric (a good road is both safe and fast). There's plenty of examples of safe and performant roads. We wouldn't want to make the 50 mile commute to the shops or work over the course of several days.

Well I don't have stats for the US, but in the UK speeding isn't the main problem. Represents about 7.5% of our accidents. Most of our accidents are right of way violations and those are no running joke. So reducing interactions between drivers and maintaining flow is key.

I think Elon said we'll definitely have that next year right?

I think a lot of these measures lull people into a false sense of security and promote a habitual lack of attention. This means if they're going to be there they need to be bullet proof. I'd personally scrap lane assist.

The attitude to driving now seems to be, press the go pedal and unless an alarm is sounding, I don't have to think. Something else will switch my wipers on and off, headlights, tell me if I'm out my lane, brake if someone is in front, beep when i'm about to hit a wall etc. so why look ...all until it doesn't or they encounter a situation there isn't a sensor for.

I don't like the disrupting the flow of traffic idea. Accidents happen in places where the flow is disrupted. E.g. junctions, give ways, reasons to stop.

It's like making a knife blunt to make it safer, in reality it makes it way more dangerous

"Head first design patterns"

I also read: - Beautiful code - Code complete - Design patterns - Refactoring - Object oriented analysis and design and many others, but honestly, the first one was a quick and easy read and gave me 80% of it

Gamification of contribution overjustifies agreement in relation to correctness. This creates a fascinatingly robust echo chamber of, in some cases, pure nonsense.

Not familiar with the area(s). But there's two I believe? One over a bridge that terminates after a roundabout the otherside, and one in Edinburgh.

How? Did machine learning leap forward yesterday when I wasn't looking? Did we achieve even basic common-sense reasoning? Or are they planning to unleash several tonnes of machinery onto the roads filled with passengers guided with little more than a CNN and a sprinkle of reinforcement learning that knows how to stay between the lines?

Great article, very well explained.