October’s DISCOVER magazine has a nice article about the Large Hadron Collider (LHC), the colossal particle accelerator which amongst many other things may reveal the Higgs Boson and the secret of gravity. The LHC is 27km long and requires a frankly ridiculous amount of power to fulfill its single goal of making particles crash at high speeds. But how much is “a frankly ridiculous amount”? What exactly is a “high speed crash” in particle physics?

‘Just the facts’ is only enough if your audience can put them in context without your help

Wikipedia’s entry, likely updated by and for professionals with more than a passing knowledge of the field puts it as follows:

The collider tunnel contains two adjacent parallel beam pipes that intersect at four points, each containing a proton beam, which travel in opposite directions around the ring. Some 1,232 dipole magnets keep the beams on their circular path, while an additional 392 quadrupole magnets are used to keep the beams focused, in order to maximize the chances of interaction between the particles in the four intersection points, where the two beams will cross. In total, over 1,600 superconducting magnets are installed, with most weighing over 27 tonnes. Approximately 96 tonnes of liquid helium is needed to keep the magnets at their operating temperature of 1.9 K, making the LHC the largest cryogenic facility in the world at liquid helium temperature. Superconducting quadrupole electromagnets are used to direct the beams to four intersection points, where interactions between accelerated protons will take place.

Once or twice a day, as the protons are accelerated from 450 GeV to 7 TeV, the field of the superconducting dipole magnets will be increased from 0.54 to 8.3 teslas (T). The protons will each have an energy of 7 TeV, giving a total collision energy of 14 TeV (2.2 ?J). At this energy the protons have a Lorentz factor of about 7,500 and move at about 99.9999991% of the speed of light. It will take less than 90 microseconds (?s) for a proton to travel once around the main ring – a speed of about 11,000 revolutions per second. Rather than continuous beams, the protons will be bunched together, into 2,808 bunches, so that interactions between the two beams will take place at discrete intervals never shorter than 25 nanoseconds (ns) apart. However it will be operated with fewer bunches when it is first commissioned, giving it a bunch crossing interval of 75 ns.

[...] While operating, the total energy stored in the magnets is 10 GJ (equivalent to 2.4 tons of TNT) and the total energy carried by the two beams reaches 724 MJ (173 kilograms of TNT).

Technical, yes. Informative? Maybe, but only if you know enough about the field to make sense of the units of measurement presented. This is probably enough for a particle physicist. It likely doesn’t generate understanding in a layman.

If you need to, give useful contexts to the facts

DISCOVER’s Lisa Randall goes one step further in an attempt to translate for the reader:

“I learned more about the backstory [of the Large Hadron Collider] during my visit. Keep in mind that the ultimate goal for collisions is a center of mass energy of 14 TeV, or trillion electron volts. I realise these might be unfamiliar units, so to give some perspective, it is seven times the energy of the Tevatron particle accelerator at Fermilab in Illinois, which is presently the highest-energy machine, and 15,000 times the energy contained in the mass of a single proton at rest” Lisa Randall, ‘The Heart of the Matter’

The effort is laudable, but it falls short of really communicating what 14 trillion electron volts because it uses examples that are only meaningful to the kind of reader who most likely already understands that unit of measurement. The problem is that electron volts, Fermilab and the energy of a stationary proton are all part of the same language.

Know what language your audience is speaking

Description is not communication. You can’t deposit knowledge in a person’s head, just as you can’t stick new leaves directly onto a plant. Successful communication is about feeding your audience the right blend of facts, stories, examples and experiences so that their understanding of a topic can grow within what they already know.

In both cases, the comparisons given (“largest cryogenic facility”, “99.9999991% of the speed of light”, “seven times Fermilab”, “mass of a singe proton”) will not mean much without similar facts in their mind acting as ‘hooks’. But what if you put it in terms that made comparison easier?

• At their top speed, these protons will travel the 27km circuit (about 10 laps of a Nascar oval, or two thirds of a marathon), 3,500 in the time it takes to blink an eye
• Each of the 1,600 superconducting magnets arranged around the track (one every fifty feet on average) weighs about the same as two fully laden container trucks
• It takes 120 MW to run the LCH, approximately the power consumption of all the Canton State of Geneva where the LHC is located. That’s half a million people, about the population of Atlanta.

This kind of translation isn’t always necessary, but always make sure to speak the same language as your audience. It’s not what you say that matters, it’s what they understand.

The other day I popped into my local branch of Lloyds TSB and saw this next to the ATM:

Taking a closer look:

A heatmap shows the busiest times for that particular branch, with a bit of analysis above to help people make sense of it and explain, for example, the grey block on wednesday morning (staff training day).

Short interview with the guys who put it up

This is very interesting – a bank visualising its data to change customer behaviour. Rebecca Reeves, the branch manager, was kind enough to answer a few questions about it:

Raphael D’Amico: Thanks for agreeing to explain this display a little bit. So, why did this start?

Rebecca Reeves: We noticed that we had both business and personal account holders coming in the lunchtime rush hour, even though business customers can generally choose to come at any time of the day. The idea behind this was to try to get our business customers to come by when the branch was quieter.

RD: How does it work?

RR: It uses the transactions done in the branch. We record the data and a team at the head office feeds back these heatmaps.

RD: Is it just this branch?

RR: No, it is done across the country.

RD: Has it worked?

RR: It has actually. We started recording data about a year ago, and put the first heatmap on the wall six months ago. When we analysed the data again quite recently we saw that customer transactions were more spread out across the day.

In particular, it didn’t make that much of a difference to personal customers – they still came mostly at lunchtimes – but business customers did start coming more often at other times.

RD: How did you measure the improvement? Did you measure queue lengths, for example?

RR: Just by sight – the only formal measurement was the transaction data, which tells us the time and type of customer, for example.

RD: Thanks for your time.

Neat, but what could it do better?

This idea is clearly a good one and has worked, but there is as always room for iteration. Here are a few suggestions:

• Make it bigger and move it slightly further from the cashpoint. Putting it next to a cashpoint is a good idea and gives it exposure, but the size and positioning means you have to be close to the wall to see it. This leads to two less than ideal situations:
  1) You look at it while taking cash out, which slows you down and makes people behind you wait.
  2) You take a proper look afterwards, which means you have to stand directly next to and very close to the next person taking money out, which tends to make both you and them uncomfortable. It is almost a social taboo to do this, and probably keeps a a fair few people away.
  A larger, more legible display would solve this.
• Put it near to other queues in the branch, not just the ATM. The queue for the bank teller is longer than the one for the ATM, which would make customers even more receptive to this kind of display.
• Measure how long you are actually taking to serve customers. While the transaction data is a good proxy, Lloyds should spot check exactly how long it takes them to serve each customer (how do they promise four minutes?). This may also allow them to segment their customers better – perhaps there are some transactions that are more time consuming and could be addressed in the heatmap display.
• Show customers the changes. Showing people that this display has already changed behaviour may make it even more effective through social proof.
• Share data. Comparing customer patterns across branches might reveal some good techniques they can learn from each other. I didn’t ask about this, so it could be that the branches already do this – I imagine the data analysis is centralised for this purpose.

It’s really great to see a large organisation using this kind of technique (particularly a bank, right now!).

Are there other companies feeding the behaviour of their customers back to them?