Readers’ comments
Your views and feedback from email and the web
Moore’s outlaw
I’ve read that we’re rapidly reaching the end of the line in terms of shrinking semiconductors. We hit the clock speed limit around 15 years ago so manufacturers are running out of options to deliver further gains in performance or efficiency. Yes, you can keep adding cores to processors but a lot of tasks cannot simply be divided into dozens of smaller jobs performed in parallel.
It’s said the only way forward is moving from largely generic CPUs to processors optimised for specific tasks. Apple’s move to designing its own silicon, coupling hardware and software, looks like a very shrewd move at this point. Google is following suit, designing its own chips, so I wonder what the “Wintel” platform can do to avoid stagnation and decline.
Personally, I’m looking forward to a new era where software and web developers have to focus on efficiency, making the most of what the laws of physics allow, rather than relying on Moore’s law to bail them out. Lawrence Hassall
Totally wired
I’ve now managed to use nearly half a mile of Cat6 cabling since the lockdown started, having been called up by family, friends and friends of friends to sort out Wi-Fi problems so they can work at home. Every time, it’s been that the Wi-Fi signal is too weak or unreliable to reach where they’ve put their home offices.
Each time, I asked if it’ll be a permanent fixture at home with most saying “yes” so I pushed them down the route of using cabling direct from router to computer, hence the half-mile of cable. However, others wanted much better Wi-Fi so, in these cases, I supplied them with DrayTek extenders (I’d won a box of 15 of them at an auction for £40), which were placed near to their working areas.
Some of those with weak Wi-Fi signals had mesh systems already in place, but these were struggling due to things such as thick walls and even foil-covered insulation, which meant the mesh systems were only repeating a weak signal. You really should have a mesh system made up of wired range extenders.
If you’re renovating a house, think about your home computer network in the same way as you would the electrical sockets and lights because it should be considered part of the house’s infrastructure. This was hammered home by a mate who was renovating and enlarging his house for his family and asked me ten years ago how to future-proof the house. I told him that TV will come via the internet so some hard cabling would be ideal. He installed two cables to each of the four bedrooms and also to the main and kitchen TVs. Come lockdown, five of the family had no problems working from home in respective rooms with wired Ethernet and the ability to use the TV screens as large main/second monitors. Michael Ashworth
Making the grade
I would like to take issue with Nicole
Kobie’s article about the exams algorithm ( see issue 313, p21). Nicole is correct that exams are poor at grading “ability”, except for coping with stress and memory. Exams disadvantage students who are poor at exams and those high-achievers at teacher-assessed tasks such as research, group work, presentations and so on. It’s well established that teacher-predicted grades are inflated compared with exam grades since their assessments include some different abilities as listed above.
The exams algorithm was tasked with making the teacher grades more like previous cohorts’ exam grades, which it did – although it was still generous, meaning more students achieved their university entry grades than if they had taken exams. Overall, the algorithm benefited students. Indeed, exam boards do something similar every year, shifting all grades to make them better match previous cohorts. It’s a statutory requirement.
The algorithm grades were founded in the reality of teacher predictions from real student work and of real students cohorts’ previous performance. This is just a different reality from that of students doing examinations. Some high-achieving
students were disadvantaged, just as is the case with examinations as shown above. The point is that it was simply a different group of students who were disadvantaged and, since the algorithm grades were generous, it was a smaller group. So why is that a problem? The disadvantages could probably have been minimised by the appeals system. Appeals are always part of the process.
However, those disadvantaged or not achieving university entry grades were the focus of media attention in a manner that does not happen with examinations. Aggrieved parents and the wider community were swept along on a tide of emotion, and any rational analysis of the process was impossible. The government couldn’t risk a loss of support from the parents and the wider community that might affect future votes. It capitulated. So, in England, an overall increase of 12% in A or A* grades compared to previous cohorts is considered fair. That this disadvantages the many hundreds of thousands of students in previous and future cohorts is of no concern.
Nicole may have accurate views on the use of algorithms in general, but I think she’s incorrect in describing this as the “exams debacle”. It was a “debacle” because of the way that emotional thinking triumphed over rational thinking, which led to an adequate algorithm becoming unreasonably castigated.
Dr Ray Harper, retired lecturer
Futures editor Nicole Kobie replies: The government didn’t perform a U-turn because of over-emotional press coverage but because it couldn’t rationally defend this algorithm. Nor should you. That the existing exam system is unfair doesn’t excuse new mistakes. The annual re-shifting of grades that you describe is based on actual exam data, this wasn’t. The head of the Royal Statistical Society has said the errors in the system were foreseeable and addressable, but a lack of accountability and transparency led to a failure to fix known faults despite being spotted months in advance. Rationally, that sounds like a debacle to me.