Why would that be interesting? Well, let’s have a look at an imaginary scenario.
Imagine that you and your neighbor work at the same company. You have the same type of electric car. And atop of this, you both charge the car at the same smart chargepoint. Each on one side. Every evening, at roughly the same time – both of you park your cars and hook them up to the chargepoint.
Yet every evening you see the light on your neighbours side of the chargepoint turn green . This means it is done charging. Yours isn’t. For hours. Why is this? Why does your neighbours car take priority over yours? Is it because he is a shareholder of the company that runs the chargepoint? Is it mere chance?
You have a Chinese car. The chargepoint in your street is co-owned by an American firm. On the news, you see tensions between the Chinese and American governments rising. The Chinese government decides on new trade embargoes against America. From the following day, your car seems to charge slower. A friend isn’t able to charge his Chinese car at all due to sudden ‘maintenance issues’. Coincidence? Or a supple American attack on the Chinese reputation?
These are just illustrations. I have no indication that this is going on right now, and not in the near future. Also, these are not matters of live and death. They are nuisances. It is just about charging a car. But when you are treated this way by your thermostat, coffeemaker, chargepoint and television at the same time, it can become annoying.
Smart chargepoints
So should we refrain from these technologies at all? Yet there are good reasons for smart charging: controlling the moment on which electric cars charge allow us to store sustainable energy on a large scale, to prevent investments in electricity grids, and to lower the price we pay for driving electric cars.
So, it makes sense to charge cars when the grid is not overburdened, and the price is low, and the sun is shining. The smart charging algorithms we are working on in our pilots are programmed to optimize these various goals. They also choose between conflicting goals if need be.
But it is unlikely this is the end of optimization. In the coming years we will likely see automatic optimization on users schedules. The chargepoint will check your outlook of gmail schedule to see at what time charging really needs to be done. And it will check your next destination to see how much electricity you will need. And it will check for traffic jams, suggesting you to postpone your trip (as you will be in traffic anyhow) while at the same time profiting from cheap electricity charging.
All this occurs behind the scenes. But as a car driver, you have no easy way to check how decisions regarding to charging are actually being made. And whether these decisions are fair to you. Or whether it’s related to the trade embargo that was mentioned at the beginning of this article.
Transparency
In an ideal world, we would all be in control of the algorithms that are working for us, as idealized in Richard Stallman’s ‘Free Software’ concept. Yet as this may be too difficult to achieve. We feel transparency might be a first step towards not being maltreated by what algorithms are doing.
Of course, we have a long tradition with transparent algorithms: it’s called open source. And open source is great. But it doesn’t go far enough. For open source to be useful to understand what algorithms are doing, you have to be a programmer. And you have to be behind a computer. That basically rules out the majority of the population.
So we need a new way of thinking about transparency. What does transparency mean in an internet-of-things connected city?
We have three concepts that illustrate what transparency could look like. This first is the Chargepoint-as-a-kiosk, in which the device itself has features to increase its transparency to users. The second concept is the idea of Crowdchecking. The third concept is a digital seal, meant to inform people that a device is trustworthy.
CHARGEPOINT-AS-A-KIOSK
Information should be presented when it is relevant in order for it to be meaningful. That is why it is the chargepoint itself that informs the user about what it is doing. It does so with a screen that is fixed on the device. It would be completely transparent to see the running source code on this screen. Also, it would not make much sense to most people. But we can show what the software is doing in a more abstract way by showing what it is trying to achieve, and how it is trying to do so. For example:
“Hi, I’m an intelligent chargepoint. My job is to charge the two cars to my left and to my right. I’m trying to do this as cheaply as possible, using as much solar energy as possible, whilst not overburdening the electricity grid.
Right now, electricity is cheap, so I’d love to charge both cars as fast as possible. However, the electricity grid is nearing its maximum capacity. So I can only charge at half speed. Also, the car to the left has indicated it has to leave in a hurry so it uses most of this capacity.”
Combining this with the state actual variables the chargepoint uses in this decision ( 10 cts / kWh, cable load 90%, sunniness 30%, charging speed car 1 11kW, charging speed car 2 4 kW) a user has a fair insight in what is going on. He can understand why the chargepoint acts the way it does. Also, he might spot major aberrations. For instance the chargepoint claiming it is charging with solar energy while it is raining.
Information for non-users
When thinking about a transparent chargepoint, we initially think about transparency for the drivers that charge their cars. But, as we are talking about a chargepoint in a public street, we might consider passerby’s as well. Let’s call these non-users. Non-users might be more interested in what the device doesn’t do. For instance: although it is RFID equipped, it is programmed not to start reading RFID beacons until a button is pressed. Or: this device has a camera, but its lens cover is only opened if you are within a 1 meter – range, it stores voice data locally and deletes it after 10 minutes. These details should be iconized and clearly visible on the chargepoint.
THE THIRD PARTY APP
The second concept is a smartphone app that is issued and maintained by a third party. Not only has it a different form, this form makes it suitable for different functionalities.
The first functionality is that it can be used to compare and choose between chargepoints of different companies. Not only does it allow you to compare cost and availability, as current apps do, it allows you compare on different aspects as well. You might not want to charge at chargepoint company that is backed by Facebook or major banks or Russia or China. The app would simply hide these from your search results.
The second functionality is that the app discloses both professional reviews of the chargepoint software, as well as user experiences. A bit like the a consumers guide.
For this to work users allow their (anonymized) data to be used by the third party for review purposes. This allows both professional and amateur researchers to spot possible unfair behavior.
Let’s examine how this could work by considering the algorithms that try to sell us flight tickets. They are programmed to use all the information they can get to establish the maximum price we will pay for a ticket. There is a substantial number of people who consider this unfair. What if everybody uploaded their ticket buying history so it could be compared to others? If we’d all do this with the flight tickets we buy and some basic characteristics, it would become much clearer how the algorithms work that treat us differently on browser versions, times we visit the website before we decide to buy, etc.
This could be applied to chargepoints as well. If we would have a set of a large number of car owners, their charging preferences, car brands and their actual charging data, it would be comparatively easy to spot unfair behaviour. Like, particular brands of cars, not being charged at full speed at particular charge points, etc. Or cars of famous people or politicians taking charging priority over others.
THIRD PARTY SEAL
We could also opt for a third party solution that does not give insight in what is happening, but simply asks to trust the users when it says the particular chargepoint is ‘ok’. An objective third party reviews the software and firmware of the chargepoint. It provides a digital seal. The seal is linked to a particular sourcecode via an MD5 hash. This eliminates the risk the chargepoint software is ‘secretly’ changed after the seal has been issued.
INSIGHTS
Would any of these concepts prevent the situations described at the start of this article? Possibly. The idea of requiring smart chargepoints to explain their goal, and the way they are trying to achieve that goal seems a start. At least it offers a way to understand in general wat is going on. Or at least: supposed to be going on. Because what something is wrong somewhere deeper in the system? At that point, the Crowdchecking functionality is interesting because it gives us a fair chance to spot a situation when we are singled out and treated distinctly differently than the rest of the population. But to actually pinpoint unfair behavior to a particular part of the algorithm requires the ability to judge the actual source code, and seems beyond the ability of the user himself. For this we need independent experts. Here , the concept of a third party seal can be useful.
However, this exploration gives rise to at least three new questions.
- Will any of these concepts be effective enough to thwart a cunning and malicious actor?
- In reality, there is not ‘one algorithm’ at the heart of what a chargepoint is doing. What the chargepoint does and does not do is the result of the back and front offices of about five parties involved in smart charging a car, not counting the driver himself. So if you feel something goes wrong, who is to blame? And what information do you need to pinpoint this problem to the right party?
- It is difficult to imagine any concept without an objective trusted third party. How do you finance an objective third party?