Tag Archives: automation

Improving the accuracy of waist circumference measurement

Waistline needs to be measured for clothes size determination and health evaluation, but the instructions to do it are vague: keep the stomach neutral, neither sucked in nor bulged out, hold the measuring tape not too loosely nor too tight. A person’s waist expands and contracts with the breath, so at which point of the breath cycle should the measurement be taken? How much tension should the measuring tape have?

Averaging several measurements would be more accurate than a single vaguely defined one, both in the sense of replication across different measurers and in the sense of corresponding to the physical quantity of interest (health status, tightness of the clothes). One simple way is to average the measured perimeter of a maximally sucked-in gut and a maximally pushed-out gut.

A better measure, but quite difficult to take, is to average the waist circumference across the whole cycle of breathing in and out, for several cycles. A device to do this would be a loop of flexible measuring tape pulled into a roll by a spring, similarly to a construction measuring tape made of metal. The roll would contain electronics which continuously record how much tape is in the roll, therefore the length of the loop outside the roll. The length data could be averaged to obtain the average waistline.

A more modern method is to use two cameras pointed at the person from two sides at waist height, recording a 3D video of the waist over the breath cycle. The video could then be averaged on a computer to find the mean volume. Whole-body volume could be determined similarly – no need for Archimedes’ Eureka method of submerging the body in a bathtub with a volume scale on it. The volume and the weight determine the density of the body, which gives partial information about its fat percentage.

Computer vision training sets of photos are endogenous

In principle, every pixel could be independent of any other, so the set of possible photos is the number of pixels times the number of colours – billions at least. No training data set is large enough to cover these photo possibilities many times over, as required for statistical analysis, of which machine learning is a subfield. The problem is solved by restricting attention to a small subset of possible photos. In this case, there is a reasonable number of possible photos, which can be covered by a reasonably large training data set.

Useful photos on any topic usually contain just one main object, such as a face, with less than 100 secondary objects (furniture, clothes, equipment). There is a long right tail – some useful photos have dozens of the main object, like a group photo full of faces, but I do not know of a photo with a thousand distinguishable faces. Photos of mass events may have ten thousand people, but lack the resolution to make any face in these useful.

Only selected photos are worth analysing. Only photos sufficiently similar to these are worth putting in a computer vision training dataset. The sample selection occurs both on the input and the output side: few of the billions of pixel arrangements actually occur as photos to be classified by machine vision and most of the training photos are similar to those. There are thus fewer outputs to predict than would be generated from a uniform random distribution and more inputs close to those outputs than would occur if input data was uniform random. Both speed learning.

When photo resolution improves, more objects of interest may appear in photos without losing usefulness to blur. Then such photos become available in large numbers and are added to the datasets.

Reduce temptation by blocking images

Web shops try to tempt customers into unnecessary and even harmful purchases, including grocery and food ordering sites which promote unhealthy meals. The temptation can be reduced by blocking images on shopping websites. I find it useful when ordering food. Similarly, Facebook and news sites try to tempt viewers with clickbait and ads. To reduce my time-wasting, I make the clickbait less attractive by blocking images. The pictures in most news stories do not contribute any information – a story about a firm has a photo of the main building or logo of the firm or the face of its CEO, a “world leaders react to x” story has pictures of said leaders.

The blocking may require a browser extension (“block images”) and each browser and version has a little different steps for this.

In Chromium on 20 Jan 2021, no extension is needed:

1) click the three vertical dots at the top right,

2) click Settings to go to chrome://settings/,

3) scroll down to Site settings, click it,

4) scroll down to Images, click it.

5) Click the Add button to the right of the Block heading. A dialog pops up to enter a web address.

6) Copy the url of the site on which you want to block pictures, for example https://webshop.com into the Site field.

If seeing the images is necessary for some reason, then re-enable images on the website: follow steps 1-4 above, then click the three vertical dots under the Add button under the Block heading. A menu of three options pops up. Click the Allow option.

Alternatively, you may block all images on all websites and then allow only specific sites to show images. For this, follow steps 1-4 above, then click the blue button to the right of the Allow all (recommended) heading. Then click the Add button next to Allow. A dialog pops up to enter a web address. Copy the url of the site on which you want to block pictures, for example https://webshop.com into the Site field.

Ebay should allow conditional bids

Ebay should allow buyers to bid for a single item across multiple auctions: make a bid for one item, then if outbid, automatically make the same bid on the next identical (as defined by the buyer) item and so on. This increases efficiency by joining multiple auctions for identical items into one market with many sellers and buyers. It also reduces selling times, because a buyer who just wants one unit does not have to wait until being outbid before bidding for the next identical item. Buyers generally are not continuously watching the auction, so there is a delay between being outbid and manually making the next bid. Buyers are willing to pay to reduce the delay, as evidenced by purchases at “buy it now” prices greater than the highest bids in the auctions.

More generally, bids conditional on being outbid would help merge auctions into markets, gaining efficiency and speed. For example, a buyer has different values for used copies of the same item in different condition and wants just one copy of the item. Conditional bids allow the buyer to enter a sequence of different-sized bids, one for each copy, with each bid in the sequence conditional on the preceding bids losing.

Linking the bids is not computationally difficult because Ebay already sends an automatic email to a buyer who has been outbid. Instead of an email, the event of being outbid can be used to trigger entering a bid on the next copy of the item.

Faster selling times benefit everyone: sellers sell faster, buyers do not have to waste time checking whether they have been outbid and then making the next bid, Ebay can charge higher fees to appropriate part of the increased surplus from greater efficiency. Ebay can also use the data on which items buyers consider similar enough to classify products and remove duplicate ads.

A browser extension or app can provide the same functionality: an email with title containing “You have been outbid” triggers code that logs in the user (with the credentials saved into a password manager or the browser) and types in a bid on the next copy of the item.

Identifying useful work in large organisations by revealed preference

Some members of large organisations seemingly do work, but actually contribute negatively by wasting other people’s time. For example, by sending mass emails, adding regulations, changing things for the sake of changing them (and to pad their CV with „completed projects”) or blocking change with endless committees, consultations and discussions with stakeholders. Even if there is a small benefit from this pretend-work, it is outweighed by the cost to the organisation from the wasted hours of other members. It is difficult to distinguish such negative-value-added activity from positive contributions (being proactive and entrepreneurial, leading by example). Opinions differ on what initiatives are good or bad and how much communication or discussion is enough.
Asking others to rate the work of a person would be informative if the feedback was honest, but usually people do not want to officially criticise colleagues and are not motivated to respond thoughtfully to surveys. Selection bias is also a problem, as online ratings show – the people motivated enough to rate a product, service or person are more likely to have extreme opinions.
Modern technology offers a way to study the revealed preferences of all members of the organisation without taking any of their time. If most email recipients block a given sender, move her or his emails to junk or spend very little time reading (keeping the email open), then this suggests the emails are not particularly useful. Aggregate email activity can be tracked without violating privacy if no human sees information about any particular individual’s email filtering or junking, only about the total number of people ignoring a given sender.
Making meetings, consultations and discussions optional and providing an excuse not to attend (e.g. two voluntary meetings at the same time) similarly allows members of the organisation „vote with their feet” about which meeting they find (more) useful. This provides an honest signal, unlike politeness-constrained and time-consuming feedback.
Anonymity of surveys helps mitigate the reluctance to officially criticise colleagues, but people may not believe that anonymity will be preserved. Even with trust in the feedback mechanism, the time cost of responding may preclude serious and thoughtful answers.

Feedback requests by no-reply emails

We value your feedback” sent from a no-reply email address shows not only that the feedback is not valued, but also that the organisation is lying. More generally, when someone’s words and deeds conflict, then this is informative about his or her lack of truthfulness. If in addition the deeds are unpleasant, then this is the worst of the four possibilities (good or bad deeds combined with honest admission or lying).

The fact of sending such no-reply feedback requests suggests that either the organisations doing it are stupid, needlessly angering customers with insincere solicitations, or believe that the customers are stupid, failing to draw the statistically correct (Bayesian) conclusion about the organisation.

Some organisations send an automated feedback request by email (Mintos) or post (Yale Student Health) in response to every inquiry or interaction, even ones that clearly did not resolve the problem. The information about the non-resolution could easily be scraped from the original customer emails, without wasting anyone’s time by asking them to fill out feedback forms. The inefficient time-wasting by sending feedback requests is again informative about the organisation.

Free wifi lies

Many airports, hotels and other public places advertise free wifi, but in a significant fraction of them, the wifi does not work, e.g. in Doha Airport. One view about this is that you get what you pay for. On the other hand, the claim of providing free wifi makes people try to connect and wastes their time. Everyone would be slightly better off if the non-working wifi was not advertised – the advertiser would save the cost of printing the „Free wifi” signs and the visitors would save time.

It is cheap for the service provider to check whether the wifi is in fact working – just program a few cheap used smartphones to periodically try to connect to the wifi and send a notification to IT if the connection attempt fails. The connection failure may even trigger an automatic restart of the router.

Some airports may have wifi available, but only to a restricted group of people. For example, in India, connecting requires a local phone number, which most international travellers do not have. In Singapore and Shanghai airports, the wifi requires either a local number or scanning one’s passport in a kiosk, and the kiosks are sometimes out of order. Again, looking for the kiosk and trying to scan wastes time.

Intermittent wifi may be better or worse than none, depending on what fraction of time it is available and people’s average time cost of trying to connect.

Online check-in lies

Almost all airlines advertise the option to check in online and send email reminders to do so. In my experience, some airlines (Qantas, Air New Zealand and Qatar Airways) frequently do not allow online check-in despite falsely claiming that it is always available, or only unavailable to underage people and large groups. Email reminders to check in online seem like mockery in this case, but are still sent.

The false advertising of online check-in wastes customers’ time by encouraging them to start the data entry process. Often the process can be almost completed and only at the end does a message appear saying that online check-in is unavailable. To reduce the wasted time, the process should be stopped as soon as possible whenever it cannot be completed but is nonetheless started. It seems a simple IT fix to not send the automated reminder emails when online check-in is unavailable, and display the message „Online check-in unavailable” at the start of the data entry process instead of at the end.

A similarly ironic tone to falsely advertising online check-in is achieved by sending „we value your opinion” emails from a no-reply email address, or claiming to listen to customers but providing no contact email or phone on the website. Such mockery is practiced by many large companies. Sometimes the firms provide a feedback form that is user-unfriendly and requests lots of personal data. Or they may refer inquiries to a very limited FAQ section. The FAQ sometimes lists questions no real customer would ask, along the lines of „What makes your product so excellent?” These questions are in the FAQ just to let the company repeat their marketing slogans.

Ventilation switch that detects odours

Many homes in the US have a switch to turn on the ventilation in a toilet or bathroom. Also, all kitchen range hoods and laboratory fume hoods I have seen must be manually switched on and off. Of course the ventilation could be run continuously, but this would be noisy, waste electricity and remove warm air (or cool air-conditioned air in hot weather) from the building. For toilets, bathrooms and kitchens, the main reason to ventilate the room is only temporary – removing odour or humidity.

An untapped business opportunity is to produce a switch that detects odours or humidity and turns on the ventilation just long enough to remove these. Humidity detection is easiest – just connect a hygrometer to the switch. Detecting smelly gases such as grease vapour in the kitchen, hydrogen sulfide, methanethiol and dimethyl sulfide in the toilet may require spectrometry or a chemical reaction. Laboratory gases are probably the most difficult to automatically detect due to their variety.

Parking space availability detection from the air

Much fuel and time is wasted by drivers zigzagging through a parking lot looking for an empty spot. Some parking garages have automated systems for detecting which spaces are empty and directing drivers to these using electronic displays or coloured lights on the ceiling. Drivers of course prefer to spend less time looking for a space and to find places closer to the exit, so providing availability information gives a commercial parking building a competitive advantage and reduces traffic and pollution inside it. The same availability information would be helpful for outdoor parking lots, including the tops of garages that are open to the sky.

A business opportunity is to develop an app that uses either satellite images or cameras on drones or surrounding structures (highrises or streetlights) to detect empty parking spaces and direct the drivers to these. The cheapest black and white security camera is sufficient to distinguish a car from an empty rectangle of asphalt. Even at the busiest times, such as the peak shopping hours on Saturday afternoon, the parking availability info only needs to be updated about once every 30 seconds to be maximally useful. The only hurdle for this system is the placement of the cameras, because buying two satellite photos per minute for every parking lot may be expensive for the app provider. Drones may not be permitted over urban areas. The owners of the structures surrounding the parking lot may ask for payment to allow camera installation on their property. Of course, permissions should not be a problem if the surrounding buildings are retail establishments who profit from more customers parking closer to them, but the upper floors of tall buildings are typically offices or residential spaces.