The limits of simulations

I’ve recently been assigned to a team working on a simulator to model and test some business scenarios. The details aren’t important but the upshot is that a model will be built, some decision points assigned percentages and times, then 1,000 transactions run through to see what the baseline is. After that it’s a matter of playing with the times and percentages to see what might happen if some changes are made. As with many things, those looking for the output of the simulation have essentially decided on the scenario they think they want to use and they exercise it to see if there are any compelling reasons to reconsider.

Simulations and modelling work well for simple, easily defined processes. With any process involving people, assumptions have to be made about average times and percentages. If you have a certain time to get the work done, then there is a limit to the level of detail and complexity that can be built into a model, how good any underlying historical data is for calculations, the ability to check back for cyclical patterns and the inability to predict how things like a global financial crash will affect the current model state. In short, it’s a best guess that will not be completely accurate.

With that in mind, a US company is selling software that will read facial expressions during a job interview to analyse their performance. Hirevue claims that their software will analyse language and expressions to determine the best candidates. The process is based on an automated video interview captured by phone or computer. This is then compared to people who were found to be good at their jobs. They are given a ranking of 1-100 based on a claimed 350 factors, mostly to do with the use of language, including speed, complexity and sentence length. The claim is that the process has less bias than can be found in human interviewers. There is a long list of potential difficulties with such an approach, and, depending on the base dataset, could introduce other biases, especially if the candidates are in some way unconventional. People who would do well with a human may not do so well with a computer and so on. The approach is based on a model similar to that described earlier, with more data points and factors, but essentially it is still a best guess.

Ihaven’t written about how much of a PC you can get for a sum of money for a while now. The reason can be explained by the latest iPhone 11 Pro Max or the Samsung Galaxy Note 10. These top-end phones are expensive. The former is US$1,500 (46,000 baht) in a standard configuration, and the Note 10 is in the same ballpark. Both units are as powerful as the average home computer only a few years earlier, but you can fit them in your pocket.

Iwatched as someone asked Siri a question recently and the answer came back quickly, with relevance and with some detail. With the larger screen sizes, tablet equivalents have become all but irrelevant unless they are built into a more useful format as part of a light notebook. There are fewer and fewer homeowners buying PCs, and organisations that can afford the difference are buying notebook devices for their employees. The PC market is still good in the medium-to-large organisations, where a standard desktop approach is required and their PCs don’t need a great deal of power; hence, lower costs. Smartphones have not replaced the business alternatives, but more and more senior management are using them as standard devices, and in the end the only limit will be screen sizes.

So what will replace them? If you are a sci-fi watcher, the next step is wearable or implanted devices. This will start with basic communication technologies, probably with internal and external components. This will gradually be upgraded to the kind of phones seen in the latest version of Total Recall, once power and antenna requirements have been met. After that, I can’t even guess where it all goes.

I’ve seen the new iPhone 11 Pro Max live, and the first thing I noticed are what appeared to be large camera lenses on the three cameras on the back. Then I looked closer and the actual camera lenses were contained inside a much larger circle, so it is more of a style thing than actually required. Other makers have a large number of lenses, so once again this is an incremental catch-up rather than anything revolutionary.

In my post-Ingress Redacted time, I tried their new Prime product. I have a Samsung Galaxy 10 5G, so a fairly powerful device. On the test day, Prime locked up on me a number of times, requiring restarts, something the older version never did on my much older Galaxy S5. Things also take a lot longer to do, so Niantic has taken all the fun out of the Ingress game. I’m not sure what their long-term game plan is here.

James Hein is an IT professional of over 30 years’ standing. You can contact him at jclhein@gmail.com.