Increasingly, filtering systems are trying to keep spam out of your inbox. Sometimes, they accidentally filter out the e-mails you want to receive. To ensure important emails are not filtered into your ‘junk’ folder, you need to take action.

Unfortunately, it is not always straightforward to do this, and the process often depends on the email provider.

For some, you add the trusted domains to your Safe List. The safe list can be accessed via an ‘Options’ or ‘Settings’ link.

For others, add the trusted domains to the filter’s ‘whitelist’. You may need to search a filter’s help for how to do this — and depending on software/version, they may call a ‘whitelist’, a ‘good list’ or a similar name.

If an important email is filtered to your ‘Spam’ or ‘Junk’ folder, open the message and click on the ‘this is not spam’ link.

If an important email is still being filtered or your email client doesn’t have a ‘whitelist’, try adding the domain or the email’s ‘From’ or ‘Reply-to’ address to your address book or contact list. If this option is not available, try moving the message to your ‘inbox’, forwarding the message to yourself or replying to the email.

If subsequent messages continue to be filtered, call or e-mail your ISP’s tech support and specifically ask how you can be sure to receive all e-mail from your trusted domains. Follow their instructions for whitelisting.

It’s a shame to have to bother with all this, but unfortunately, a few unscrupulous people make it difficult for those who want to conduct business responsibly on the Internet.

Of course, making sure that you can receive emails from trusted sources is only half the story. If you send emails, there are some things you can do to reduce the chances of your emails being rejected by anti-spam software.

For more information on this, please see How to Ensure Your Emails are Received.

In today’s world of artificial intelligence (AI), where chatbots like ChatGPT and virtual assistants like Siri and Alexa have become household names, it’s easy to forget the humble beginnings of AI.

It all began with Eliza, the first chatbot, which was developed in the mid-1960s. Although basic by today’s standards, Eliza was a groundbreaking experiment that paved the way for decades of innovation in human-computer interaction.

What Is Eliza?

Eliza is a computer program created in 1966 by Joseph Weizenbaum, a computer scientist at the Massachusetts Institute of Technology (MIT). Eliza was named after Eliza Doolittle, the fictional character in George Bernard Shaw’s play Pygmalion. The program was designed to simulate human-like conversations. Eliza was capable of engaging in text-based dialogues by analysing user input and responding using simple pattern-matching techniques.

Eliza’s most famous script was the DOCTOR script, which simulated the conversational style of a Rogerian psychotherapist. The program often turned the user’s statements into reflective or open-ended questions, creating the illusion of understanding.

How Did Eliza Work?

Eliza’s functionality was based on a straightforward algorithm that identified key phrases in a user’s input and generated a predefined response. Its core relied on:

Pattern Matching
Eliza would scan the input for specific patterns or keywords, such as ‘mother’, ‘feel’ or ‘always.’

Transformation Rules
Once a keyword was detected, Eliza applied transformation rules to generate a response. For instance:

User: “I feel sad.”
Eliza: “Why do you feel sad?”

Deceptive Simplicity
By rephrasing or reflecting questions to the user, Eliza created the impression of understanding, even though it had no real comprehension.

Why Was Eliza Significant?

A New Era of AI
Eliza marked one of the first attempts to make computers ‘converse’ in a way that felt natural to users. It was a revolutionary demonstration of how computers could simulate aspects of human communication.

Psychological Impact
Users often felt emotionally connected to Eliza, even though they knew it was a machine. Weizenbaum himself was surprised by the depth of emotional responses Eliza elicited, especially from those who sought therapeutic comfort in its replies.

Human-Computer Interaction
Eliza highlighted the potential—and the limitations—of AI in mimicking human behaviour. It sparked debates about the ethical implications of creating machines capable of exploiting human emotions.

Weizenbaum’s Concerns

Although Eliza became a significant milestone in AI, Weizenbaum was deeply troubled by how people interacted with the program. He feared that users might attribute too much intelligence or empathy to machines. In his book, Computer Power and Human Reason, Weizenbaum warned against blindly trusting computers to make decisions in areas requiring genuine human judgment.

Eliza’s Legacy

Eliza’s design principles laid the foundation for future advancements in AI and natural language processing (NLP). It inspired the development of subsequent chatbots, including:

• PARRY (1972)
  Simulated a person with paranoid schizophrenia.

• A.L.I.C.E. (1995)
  Built on Eliza’s concepts with more complex algorithms.

• Modern AI
  ChatGPT and other advanced systems owe their origins to the conversational frameworks introduced by Eliza.

Even decades later, Eliza remains a fascinating reminder of how simple systems can create profound experiences.

Its influence can be seen in modern applications, from virtual therapists to customer service bots.

Eliza even outperformed an early version of ChatGPT in the Turing test, which requires a group of people to judge whether they are communicating with a human or a computer.

Conclusion

Eliza was much more than a simple program—it was a vision of what human-computer interaction could become.

While it lacked true understanding, its ability to mimic conversation demonstrated AI’s potential to engage with people in meaningful ways.

Today’s chatbots, with their vast datasets and deep learning algorithms, are far more advanced. Yet, the principles behind Eliza continue to inform and inspire the field of conversational AI.

Joseph Weizenbaum’s creation was a pioneer, sparking a technological revolution that continues to shape our digital world.

Eliza’s legacy reminds us that even the simplest innovations can leave a lasting impact.

Eliza apps

If you would like to try Eliza, you can use my online program, or if you are a paid member, install an Android or iOS web app on your phone.

Pythonista is one of the best iOS apps for coding and running Python scripts. The latest version, updated for Python 3.10, no longer supports creating a Home Screen icon for running your scripts. This can be overcome with Apple’s Shortcuts app, but it took me a while to work out how to do it…

While iOS doesn’t natively support creating shortcuts for third-party apps, you can use the Shortcuts app to create custom shortcuts that interact with Pythonista. Here’s a guide on how to create a shortcut to run a Python script in Pythonista:

1. Open the Shortcuts app on your iOS device.

2. Select “All Shortcuts”.

3. Tap the + icon to create a new shortcut.

4. Use the search bar to find the “Run Pythonista Script” action. Tap on it to add it to your shortcut.

5. In the “Run Script” action, tap “Script” and select the script you want to run from Pythonista.

6. Tap ( > ) and select “Run in Pythonista App”.

7. Tap ( v ) (at the top of the screen after “Run Pythonista Script”).

8. Select “Rename” and edit the name.

9. Select “Choose Icon”, then select a colour and an icon.

10. Select “Add to Home Screen”.

Now, you can run your shortcut by going to your Home Screen and tapping the icon you created. It will execute the Python script within Pythonista.

This doesn’t resemble a native app, but it is a convenient way of running your Python scripts. Web apps have a more native look, and I’ll be writing an article on how to create these.

PLATO (Programmed Logic for Automatic Teaching Operations) was a computer application dedicated to education and training. It was applied internationally to solve a wide range of training problems.

Typical examples of training included remedial education for basic skills, engineering logic testing, pre-simulator coaching for airline pilots, sales training and a range of office skills from typing to financial management.

The image above shows a student learning how to operate Xerox’s 9200 photocopier - a high-volume copier originally introduced in the mid-1970s as part of Xerox’s line of large duplicators. It was designed for print shops and large office reprographic departments. Nashua developed the program to teach its own salespeople.

The terminal had an interactive graphics display capability and incorporated a touch-sensitive display panel, an electronic keyboard, and built-in microprocessor logic control.

Ancillary devices such as hard copy units and audio units could be attached. The service was provided by a Cyber computer operating in Brussels, and all European users communicated with Brussels through telephone links.

Lessons could use these features in any combination. For example, a music lesson could show musical notes on the screen, play the notes on an attached audio device, and ask the student to respond to a question by touching the screen.

The screen presented information as text or line drawings. It was a high-resolution cathode-ray tube (CRT), with a viewing area of 8.5 inches by 8.5 inches (22 centimetres by 22 centimetres) on the front of the terminal.

The screen consisted of a grid of 512 dots by 512 dots (pixels), which lit in any combination to form characters, lines, curves, and figures.

The computer could fit 32 lines of alphanumeric characters on the screen with 64 characters per line.

When the user pressed a character key on the keyboard, the computer processed the individual keypress and displayed an appropriate character, as determined by the current lesson.

The terminal displayed 170 characters per second, erased 600 characters per second, or the whole screen at once. The terminal also drew and erased lines at speeds up to 60 lines per second.

The primary input to the PLATO system was the PLATO keyboard. The keyboard had character keys, much like a typewriter, and it had function keys, some of which affected lesson execution and some of which modified the character keys.

For example, the student could often receive help in a lesson by using the HELP key, or a mathematics student could type numbers with exponents (superscript) by using the SUPER key.

By using the SHIFT key, the uppercase character keys and the uppercase function keys became available. In addition to these, other characters and functions were available with the ACCESS key.

The application had two major features: computer-assisted instruction (CAI) and computer-managed Instruction (CMI).

CAI provided high-quality individualised instruction directly through the terminal. Its easy-to-use, yet powerful authoring language offered the course designer (who needed no previous computer experience) sophisticated graphics, gave the student supporting feedback, and a range of different methods for interaction, including response by simply touching the screen.

CMI provided a comprehensive instructional control system designed to maximise the effectiveness of individual learning and the measurement/enhancement of the effectiveness of training courses.

Sophisticated testing and recording capabilities allowed the system to assess individual students and assign the most suitable learning resources. It also provided instructors and training managers with analysis of individual or group performance profiles and feedback on the course’s effectiveness.

CMI designed the system to be instructor-independent, allowing students to train efficiently at any time on their own. At the same time, it gave the instructor the earliest possible identification of weak students or weak course material.

The computerised marking and record keeping freed expert instructors from unproductive clerical tasks, and, at the same time, the resulting database could be incorporated into overall assessment schemes or function as a skills inventory for Manpower Planning.

Module Management

This flowchart illustrates how the same objectives can be achieved with different media.

PLATO’s History

1959-1960

A research project was initiated in 1959 at the Co-ordinated Science Laboratory, University of Illinois, Urbana, Illinois, under the direction of Dr Donald L. Bitzer, to investigate the use of a computer to assist instruction in technical subjects.

1961–1964

PLATO ran on Ivy computers, which had very limited terminals. It focused on programmed learning and basic instructional exercises.

1965

PLATO transitioned to Control Data Corporation’s mainframes, enabling increased user access and improved graphics.

Late 1960s

The University of Illinois organised the Computer-Based Education Research Laboratory (CERL), and the PLATO project moved to a new laboratory.

Plasma display terminals were introduced (early touchscreen technology), and bitmapped graphics enabled diagrams, charts, and interactive lessons.

In 1967, Control Data and CERL had engaged in a joint development effort which produced the first economical, large-scale, computer-based education system.

1970s

The TUTOR Language, originated by Paul Tenczar and other CERL staff members, had evolved to the CONTROL DATA PLATO author language, which supported the production of PLATO lessons.

PLATO became widely installed in schools, universities, and military training centres.

Networking of terminals allowed multiple users to connect simultaneously.

Early forms of online communication included notes, files (message boards), chat, and email.

In 1973, multiplayer games were developed on PLATO, e.g., Empire and Airfight.

Early concepts of online multiplayer and social interaction.

In the late 1970s, PLATO reached its peak with over 100,000 users connected worldwide. It was also a major platform for educational software and research in human-computer interaction.

1980s

PLATO’s decline began because of:

• High costs of mainframes and terminals.

• Personal computers became cheaper and more versatile.

• Many institutions migrated away from PLATO.

Legacy and Influence

The PLATO systems were largely retired during the 1900s and 2000s, but their concepts, decades ahead of the web, influenced :

• Online forums.

• Instant messaging.

• Email.

• Multiplayer online games.

• Social computing.

• Networked learning.

• e-learning platforms.

The Moon must have made an incredible impression on the first members of the human race. It was the brightest object in the night sky, constantly and regularly changing shape.

The impression that the moon had on our ancestors may have led to the invention of the first calendars. What could be more natural than saying, “Let’s meet again at the next full Moon.”

The Moon’s phases were visible to everyone, and their use would avoid the counting and recording of the number of days since the last meeting.

It was inevitable that the month would become an inseparable part of our calendar — along with the day — marked by the movement of the Sun — and the year, which was marked by the passing of the seasons.

Synodic Month
The synodic month (or lunation) is the average period of the Moon’s orbit with respect to the line joining the Sun and Earth.

Sidereal Month
The sidereal month is the time it takes the Moon to return to a similar position among the stars.

Ecliptic
The ecliptic is an imaginary circle in the sky that represents the apparent path of the Sun across the celestial sphere over the course of a year, as viewed from Earth.

Magnitude
An object with a magnitude of 1 is considered to be very bright. It is about 2.512 times brighter than an object with a magnitude of 2. The Sun’s magnitude is approximately -26.74 in the visible band of light.

Optical Libration
Optical Libration is a phenomenon that occurs in the Moon's motion as seen from Earth. It refers to the apparent rocking or oscillation of the Moon’s surface back and forth in an east-west /north-south direction.

How many Lunar Months are there in a Year?

A natural next step would have been to link the lunar month with the year. This is where the difficulties began.

Twelve lunar months, with four seasons of three lunar months, seems a good approximation to the year. Unfortunately, the lunar month is just over 29½ days long, which means that twelve lunar months fall short of a year by about eleven days.

Starting a year eleven days early might not seem much, but after only three years, the lunar year would be a calendar month out of step with the seasons, which would certainly cause problems with the sowing of crops if you live in an agricultural community.

So, how about a year of 13 lunar months? Unfortunately, this results in a year that would be over 18 days too long.

The Metonic Cycle

In 433 B.C. Meton discovered that 19 years, or 6939.689 days, is almost precisely the same length as 235 lunar months, which is 6939.602 days. This produces a 19-year cycle comprising 12 years of 12 lunar months and 7 years of 13 lunar months. This keeps the lunar months in step with the seasons. These dates were inscribed in gold lettering on public monuments. For this reason, the number of years in a Metonic Cycle is called the Golden Number.

The first year of a cycle can be chosen arbitrarily, but the cycles now in use can be traced back to 1 BC, where the cycle begins at 1 when the New Moon occurs on 1 January (1995, 2014, 2033). You can also see whether a year is the beginning of a cycle by dividing it by 19 and checking whether the remainder is 0.

The Hebrew Calendar

In Israel, the Hebrew Calendar is used for religious purposes. It provides a time frame for agriculture and is an official calendar for civil holidays, alongside the Gregorian Calendar.

Like the modern Islamic calendar, the timing of the months in the early forms of the Hebrew calendar depended on actual sightings of the Crescent Moon.

However, this practice was gradually changed, and by 1178 AD, the calculation of the beginning of a new calendar month had been fully replaced by a mathematical approximation.

The Hebrew lunar year is about 11 days shorter than the solar year and uses the 19-year Metonic Cycle (see below) to bring it into line with the solar year. Even with this correction, the average Hebrew Calendar year is about 6 minutes and 40 seconds longer than the Gregorian calendar. This means that every 216 years, the Hebrew calendar will fall a day behind the Gregorian calendar.

The Islamic Calendar

The Muslim calendar uses a system based on lunar months with alternating months of 29 and 30 days. This means that all the months, including the ninth month of Ramadan — the month of fasting — continually cycle through the seasons.

To keep the months in step with the Moon’s phases, a further correction is made by varying the length of the twelfth month over a 30-year cycle: 29 days long for 19 of the years and 30 days long for the remaining 11 years.

Predicting the Phases of the Moon

Calculating the date of many religious festivals, including Easter, requires you to be able to predict the dates of the full Moon.

As the mean time between the Moon’s phases is just over 29½ days, you could roughly predict the next few occurrences of the same phase by alternating gaps of 29 and 30 days.

Unfortunately, this method will become inaccurate after a few months because the Moon is subject to over 1500 influences, which can affect the date of the full Moon from month to month.

You could also use a more accurate figure, such as the Synodic Mean (29.530558 days), for the lunar month.

The Metonic cycle described above is more accurate, and the table shown below gives you the dates of the full Moon for 2014–2051 with the Golden Number for each of those years.

Using the Metonic Cycle doesn’t always give the correct date for the full moon, so I’ve created a Python script that provides more accurate dates.

Paid subscribers can download this script from the Programs section of the Learning Pages Digital Vault.

Additions to Notes

In addition to quotes and notifications, I have added learning, earning, writing, productivity, finance and wellbeing tips to my Substack Notes.
View Notes

Free Business Jargon Generator app

Never be lost for words in a business meeting with this free macOS Business Jargon Generator.

Download the app and installation guide by visiting the Contents Page of my Digital Vault, scrolling down to the macOS apps section, and following the links.

Currency Conversion

Before deciding on new subscription charges, I wanted to see what they would be in different currencies. I started by doing approximations in my head, but soon became confused. Using a currency converter was easier, but I wanted to make ten conversions simultaneously.
Read more

Why I’m Reducing My Subscribers by 50%

My email open rate is averaging only 12%. Surely my emails aren’t that dull? So I decided to investigate. If you are reading this, you will remain one of my subscribers.
Read more

Effective Reading

Effective reading is an essential skill. It’s not only about reading faster; it’s important to decide what to read. It’s crucial to assimilate written information while grasping its meaning. It takes a lot of practice, but it reaps its own rewards.
Read more

Day by day, as a busy schedule consumes more and more of your time, a pile of reading material steadily mounts up. This growing pile may contain, for example:

• internal and external reports,

• newspaper and magazine articles,

• technical update material,

• memos,

• emails to be answered.

You may feel increasingly guilty because these remain unread, and that may not be keeping up to date or doing your job properly. Yet even if you could find time to cope, you know that you could not absorb it all.

Reading effectively is an essential skill, and it is not simply a matter of reading faster. It is important to understand what you decide to read.

Effective reading means assimilating written information quickly yet comprehending the essence of the printed word. It takes a lot of practice, but it reaps its own rewards.

Studies of fast and slow readers have isolated the faults which limit the reading speed of millions of people to 200 words per minute. Some of these factors that contribute to slower reading are:

• Bad reading habits

• Poor eyesight

• Bad lighting

• Author’s style

• Complexity of material

• Hard-to-read typeface

• Poor layout and presentation

• Unfamiliar words and expressions

Some of these factors may be out of our control, but bad habits are one area where slow readers could profitably change out of bottom gear - if only they knew how.

Adopt a positive attitude - you must feel that you want to do better and think that you can. Don’t seek overnight improvement, but do practise regularly.

Above all, accept that you cannot read everything. Constantly remind yourself to be ruthlessly selective. This may mean not being technically up to date in all areas of your business. Ask yourself regularly where it is essential to stay up to date.

Categorising reading matter

Try to sort your daily reading as it arrives: decide what your priorities are and then read. Take notes if necessary and digest the information and ideas. Don’t give up - the longer you persevere, the more you benefit.

You should never allow reading material to accumulate but deal with each piece as it arrives. Focus on what you need to know and then ask: is it important or urgent, or can it be dealt with later?

Important material demands that you find time to give it proper attention. Urgent but unimportant material needs to be dealt with quickly but do not spend a lot of time or energy.

Whatever the material, you must select what is essential and skim what is not; know what to look for: concentrate in bursts and save your energy for more important material.

Improving your reading

Once you have decided what you are going to read:

• Read the headings or index and get an idea of the essence.

• Skim the whole before reading any particular piece first.

• Review the conclusions or summary before starting on the detail.

• Read the introduction carefully: it may indicate which parts of the detail to avoid.

• Use a visual guide (i.e. a pencil, piece of paper or cursor).

• Annotate or make notes as you go. Do not wait for a second detailed reading - you may well forget what your first thoughts were.

Avoid the following bad habits:

• Reading aloud to yourself - don’t vocalise the words, let the words talk to you.

• Even if you are not reading out loud, try not to sub-vocalise (pronouncing each word in your mind).

• Lack of anticipation - don’t rely on every word to give you the sense of what is being said: anticipate as much as possible. Remember that the meaning of a sentence often comes at the end of the sentence.

• Reading the same sentence or paragraph over and over, you may legitimately discard any piece that holds no interest for you. But if there is something in the words you want to understand or absorb, by all means dwell on it.

• Reading word for word - don’t fix your eyes on one word at a time - let your sight move along the words at a constant rate, and take in several words at a glance.

Let’s prove that you can take a wider view and see several words at once. For example, if you can read the word...

NEVERTHELESS

…in a single glance, then you can easily read the three words...

NEVER THE LESS, in a single glance.

If you can read the following twelve letters in one quick look:

CONSEQUENTLY

Then you can certainly read the following nine letters just as quickly:

THAT IS WHY

Now try the following:

Practise using a wider eye span as much as possible. After a while, you will be able to fix your eyes in the centre of a column and read straight down the column instead of across it.

Useful hints

In addition to this general approach, there are several other useful things that you can do to help you with your general reading:

• Before any reading, do a quick ‘dash’ over the page. This will both briefly accustom your eyes to the higher speeds and will ensure that they are ‘warmed up’ for the reading task ahead.

• For absorbing what you are reading, it is better to have a slightly cool rather than a slightly warm temperature. 18°C (65°F) is ideal.

• Ensure your surroundings are pleasant, but not too comfortable! Your work area should be free of anything that could distract you while you are reading.

• During extended reading periods, make sure that you regularly look up and focus on distant objects. This relaxes and rests your eye muscles, enabling you to carry on reading without unnecessary fatigue.

• Occasionally stop, close your eyes, and try to reproduce, in your mind’s eye, a page of print, a photograph or a graph that you have just looked at. Apart from providing a useful rest, this will significantly help increase your recall while training your imagination.

• Your posture should be reasonably upright, with the distance between your eyes, and your reading material should be between 45 and 60 centimetres (15 and 24 inches).

Ask questions as you read and answer them yourself - for example:

• What is my purpose in reading this?

• What am I looking for?

• What is it about?

• What is the detail saying?

• What are the main ideas?

• How significant is it to me?

• How can I use what I am absorbing to do my job better?

At first, it will be difficult to read and question simultaneously; with practice, the dual activity will become a habit. Keep a positive attitude and a flexible approach - information and ideas are not readily communicated to a negative or fixed mind.

Effective reading is only one side of good communication. Effective writing is the other. Complex style, poor layout and unfamiliar jargon all conspire to ensure that communications will be poorly received.

And, remember, no matter how effective the reading, the recipient needs to be persuaded to the time and desire to read what has been written.

My email open rate is averaging only 12%. Surely my emails aren’t that dull? So I decided to investigate and here’s what I found.

First, I came up with some ideas for possible causes:

• Legacy subscribers

• Old/dead addresses (dropped emails)

• Attention-seeking subscribers

Legacy subscribers

I was sure that the main reason for the low open rate was that I imported subscribers from previous mailing lists.

Cleaning legacy subscribers is one of the most effective ways to raise open rates and improve deliverability on Substack.

Deleting legacy subscribers helps and doesn’t hurt.

As a result, there will be:

• Increased reported open percentage,

• Improved actual inbox placement

• Reduced future dropped emails

Optimise for audience quality, not raw subscriber count.

Old/dead addresses

When people change their email addresses without updating their subscription details, your emails will bounce.

These subscribers should be pruned because:

• Dropped emails damage the sender’s reputation

• A poor reputation causes more emails to land in spam

• This results in fewer opens even among good subscribers

Attention-seeking subscribers

A few subscribers subscribe to your publication, hoping you’ll subscribe back.

These subscribers will have a very low or zero open rate and will lower your overall open rate.

Defining criteria

My next step was to define the criteria for removing subscribers:

• Category 1 - Delete Immediately

• Category 2 - Legacy Pruning

Category 1 - Delete Immediately

I used four criteria for this category:

1. Emails dropped > 0

2. Emails opened (6m*) = 0

3. Activity = 0

4. Free subscribers

If a subscriber meets all four criteria, these addresses will no longer receive your emails and are certainly dead.

*The number of emails opened over a specific period.

This period should be adjusted depending on how frequently you send emails. My recommendation is:

• If emails are sent daily, change Critera 2 to ‘Emails opened (7d) = 0

• If emails are sent weekly, change Critera 2 to ‘Emails opened (1 month) = 0

• If emails are sent monthly, leave Criteria 2 as it is

Category 2 - Legacy Pruning

• Subscribed > 6–12 months ago

• Emails opened (30d) = 0

• Activity = 0

• Free

This high-confidence category should safely remove inactive legacy subscribers without reducing the number of loyal subscribers.

Last Chance

Before deleting, you can send a short re-engagement post:

“If you want to keep receiving this, just open this email.”

Then:

Delete everyone who doesn’t open within 7–10 days

This maximises fairness and metrics — but it’s optional.

Don’t worry about

• Losing ‘future readers’

• Hurting Substack growth signals

• Paid subscriber loss (these criteria only eliminate free subscribers)

Substack’s algorithm and deliverability systems strongly favour:

Smaller, engaged lists over large, silent ones

Working with the data

The next step is to apply these criteria to your Substack data. You can find these by going to your Substack Home page and clicking ‘Subscribers’, which you’ll find under ‘Audience’.

Not all the data is shown here, but you can add columns by clicking ‘Display’, selecting additional data and clicking ‘Save’.

Some columns can be ordered by clicking the header of a selected column.

I found it overwhelming trying to analyse all this data, so I decided to use a Python script to do the hard work for me.

First, I needed to download a CSV file of all the subscriber data.

I did this by clicking the menu (three dots on the top right), followed by ‘Export’, selecting ‘Export all columns’ and clicking ‘Export’.

I renamed the CSV file to ‘subscribers.csv’, moved it to the same folder as the SubscriberAnalysis.py script and ran the script:

After entering D, W or M and clicking ‘Enter’, the script produced a list of Category 1 and Category Subscribers.

Paid subscribers can download this script from the Programs section of the Learning Pages Digital Vault.

Associate Members can request tailored versions of this script.

Subscribe now

Removing Subscribers

The last step is to remove subscribers who meet your deletion criteria.

You can do this by returning to the Subscribers page.

Paste the subscriber’s email address in the search box at the top of the page, press Enter, select ‘Remove from list’ and click ‘Remove’.

I recently changed my subscription plans. Before I decided on the new charges, I wanted to see what they would be in different currencies.

I started by doing approximations in my head, but I soon became confused.

Using a currency converter was easier, but I wanted to make ten conversions simultaneously.

It was then that I had the idea of using a spreadsheet:

Paid subscribers can download this spreadsheet from the Spreadsheets section of the Learning Pages Digital Vault.

1. Enter the first and second currencies.

2. Then enter the current conversion rate.

Don’t enter zero or leave this cell blank.

You can find free currency converters online, or if you are a paid subscriber, you can download a Python Currency Converter script or macOS app from the Programs section of the Learning Pages Digital Vault.

It’s easy to multiply the currencies by ten to give you more options.

The last row of the spreadsheet allows you to convert custom amounts.

Paid subscribers can download Python Currency Comparisons and Currency Conversion scripts from the Programs section of the Learning Pages Digital Vault.

Associate (Foundation) Members can request tailored versions of these digital products.

Thanks to and for suggesting that I reassess my subscription plans.

Lumo is an encrypted version of AI. I heard about it because I use Proton Mail. Proton AG, a Swiss technology company, owns Proton Mail.

Proton Foundation, a non-profit organisation, has been the majority shareholder of Proton AG since June 2024. This change was designed to ensure Proton remains focused on privacy rather than profit.

Some remaining shares are held by Proton employees, a Swiss nonprofit foundation (FONGIT), and a small portion by users

What Proton AG says about Lumo AI

An AI assistant should empower you, not exploit you for your data. That’s why we built Lumo: To bring you all the benefits of AI, without compromising your privacy and data security.

AIs from Big Tech are built on harvesting your data. But Lumo is different. It was created by the scientists behind innovative, privacy-first services like Proton Mail and Proton VPN. And it’s owned by a Swiss nonprofit, whose mission is to advance privacy and never make money from user data.

We keep no logs of what you ask, or what Lumo replies. Your chats can’t be seen, shared, or used to profile you. Lumo saves nothing on our servers.

Thanks to zero-access encryption, your saved conversations can only be decoded and read on your device. Neither Proton nor anyone else can see them.

Lumo’s code is fully open source, so anyone can independently verify that it’s private and secure—and that we never use your data to train the model.

Your data isn’t shared with advertisers, governments, or anyone else. Lumo is a European service subject to GDPR, so you can delete your data anytime.

My Experience of using Lumo

Basic Use

Lumo provides similar responses to ChatGPT when asked the same question.

I asked both Lumo and ChatGPT the same question about software for a device I had just bought. Initially, ChatGPT gave a more detailed response, but Lumo gave a similar response when I toggled ‘Web Search’.

Encryption and Privacy

Lumo AI was designed to be fully encrypted, and this works very well.

Because Lumo doesn't keep information about your interests, it doesn’t tailor future responses to match your needs. ChatGPT asks your permission before doing this.

Citations

ChatGPT provides citations for its sources. Lumo provides citations when ‘Web Search’ has been activated.

Images

I wanted a cover picture for this article, so I Lumo asked to generate a 16:9 aspect ratio image of Lumo AI. This is the response that I got:

“I can’t create images directly, but you can easily generate a 16 : 9 ratio illustration of Lumo AI using a graphic‑design or AI‑image tool.”

Instead, Lumo provided me with a prompt, which I admit was better than my attempt.

Create a sleek 16:9 illustration of Lumo AI, a privacy‑first virtual assistant. Use Proton’s blue‑teal color scheme, incorporate a glowing light‑burst logo, subtle shield shapes, and modern chat bubble icons. Clean, minimalist style with a subtle gradient background.

I then used this prompt with ChatGPT, which generated the cover picture for this post.

Lumo Plus

The free version has recently become too restricted, with access limited to only the previous seven days of chats.

When I clicked the search icon, a sidebar appeared, and this notification was near the bottom:

I followed the ‘add Lumo Plus’ link, and these options appeared:

Not a mention of Lumo Plus!

So, I clicked ‘Explore other Proton plans’. This time, the page provided information about the Lumo Plus plan.

Price
$12.99 per month 
$9.99 per month if you pay annually (about 23 % savings)

• Unlimited daily chats

• Web search access

• Full chat history with search

• Unlimited favourites for quick access

• Upload and query multiple large files

• Access to advanced AI models

The Lumo Plus plan also includes selected benefits from:

• Proton Wallet

• Proton Mail

• Proton Calendar

• Proton Drive

• Proton VPN

• Proton Pass

I won’t go into these additional benefits here because this article is concentrating on Lumo AI.

Like other Proton products, the subscription plan is far too complicated.

Conclusion

• Lumo AI was designed to be fully encrypted, and this works very well.

• It’s similar to ChatGPT when it provides basic responses.

• It can’t generate images.

• The free version is too restrictive, particularly with access limited to the previous seven days of chats.

• Like other Proton products, the subscription plan is too complicated.

Thanks to for suggesting that I write this article.

Before I start, I would like to say that there is a need for identity verification to prevent money laundering and fraud and to protect the young. But, recently, I have been getting the feeling that identity verification is becoming more intense and requiring more photo ID.

I'll be sharing some of my insights on:

• Elections

• Financial Accounts

• Travel

• Buying and Selling Property

• Alcohol

• Visitor attractions

Elections

In 2022, legislation was introduced requiring voters in Great Britain* to show photo ID before voting at polling stations. This identity requirement was aimed at ‘reducing electoral fraud’, despite there being little evidence supporting this claim.

The real irony is that in our recent local elections, Boris Johnson, who was the Prime Minister when this law was introduced, was turned away from a polling station because he had forgotten to bring his photo ID.

Refusals are not recorded when photo IDs are checked outside the polling station, so any statistics about the number of people who weren't allowed to vote aren't accurate.

Previously, people who were eligible to vote received a polling card at their home address. Taking this card with them to the polling station allowed them to vote.

Anyway, I have been able to vote without producing a photo ID. In fact, I don't even have to go to a polling station! I achieved this by applying for a postal vote.

Financial accounts

Bank Accounts

Understandably, banks need to have identity verification to prevent money laundering.

Some banks have started introducing their own face-recognition systems to authorise payments - even if your phone has biometric security.

At least one bank is now asking you to record and send a video of yourself speaking certain phrases. This can be particularly frustrating if you are in a low-signal area.

Crypto Accounts

YouHodler recently refused my verification. First, it couldn't read my driver's licence (I never had problems with it being read before).

Then I tried my passport. My identity verification was refused because I was "underage"!!!

YouHodler said my identity verification was refused for "internal compliance" reasons when I questioned this.

Other crypto exchanges and wallet sites have accepted me.

Travel

Travel ID has become more of a hassle, whether boarding a ship or plane or entering and exiting a country.

Boarding a Ship

I had a ridiculous example of an identification hassle while boarding a ship in the UK. I had to show my boarding pass four times between the quayside and the ship! I had to show it to:

1. Security on the quay,

2. A crew member at the bottom of the gangway,

3. A crew member at the top of the gangway and

4. A crew member inside the ship.

New EU Regulations

The EU Entry/Exit System (EES) is an automated system for registering travellers from non-EU countries every time they cross a border into or out of the EU Schengen countries**. Before crossing the border, travellers will need to scan their passports or other travel documents at an automated self-service kiosk.

The system registers the person’s name, type of travel document, biometric data (fingerprints and facial images), and the date and place of entry and exit.

Buying and Selling Property

It's understandable that proof of identity is required when buying a property to prevent money laundering. It used to be that only your solicitor (real estate lawyer) needed proof of identity, but now estate agents (realtors) require it as well.

Alcohol

An eighteen-year-old walked into a store to buy a bottle of alcohol-free beer. Unfortunately, this "teenager" looked young for their age, so they had to provide "proof of identity".

Visitor Attractions

A couple of years ago, I bought a membership to one of my favourite visitor attractions. To gain entry, I only had to scan my membership card.

Recently, the attraction insisted on photographing members to prevent card theft. Now, every time you visit, your face is compared to the record on a phone app.

2026 Updates

Substack Chat

Recently, Substack Chat restricted UK citizens’ access to Chats because of the Online Safety Act.

This requires you to use your camera to take a photo of the front of your face, with small turns to the left and right. A proof-of-age document and a photo ID are not needed. So, this is only age verification - not identity verification.

Australia

Australia has banned under-16s from social media, but they can still access gambling sites.

EU Entry/Exit System (EES)

After many delays, the EU Entry/Exit System (EES) was implemented in October 2025 checking 5% of entries. This is increasing to 10% in January 2026 and 100% in May 2026.

UK Identity ‘Cards’

The UK is planning to introduce digital identity cards. Not everyone will have to have one, but they will be necessary for working in the UK.

USA Social Media History

Tourists and business travellers entering the USA will need to provide their social media history for the last five years. Although this isn’t strictly identity verification, it is providing more identity information.

*Great Britain includes England, Wales and Scotland, but not Northern Ireland. Hence, the term "The United Kingdom of Great Britain and Northern Ireland".

** Schengen Area Countries (as of 2025)

EU Member States in the Schengen Area

1. Austria

2. Belgium

3. Croatia

4. Czech Republic

5. Denmark

6. Estonia

7. Finland

8. France

9. Germany

10. Greece

11. Hungary

12. Italy

13. Latvia

14. Lithuania

15. Luxembourg

16. Malta

17. Netherlands

18. Poland

19. Portugal

20. Slovakia

21. Slovenia

22. Spain

23. Sweden

Non-EU Countries in the Schengen Area

1. Iceland

2. Liechtenstein

3. Norway

4. Switzerland

Partial Schengen Members

Bulgaria and Romania partially joined the Schengen Area on 31 March 2024— air and sea border controls were lifted, but land border controls remain in place for now.

Not in the Schengen Area (even if in the EU)

• Ireland (opt-out; maintains its own border controls)

• Cyprus (EU member but not yet part of Schengen)

Substack Notes

I have started posting on Substack Notes. There will be a daily quote and tip for learning, earning, writing or productivity.

Subscription Changes

The price of a Learning Pages subscription hasn’t changed for over five years. So, I’ll be increasing the price in two weeks (13 January).

I am giving two weeks’ notice so that my loyal free subscribers have the chance to upgrade at the current price.

The 25% reduction for your first paid year will be available before and after the increase.

Upgrade and get a 25% reduction

Existing paid subscribers won’t see their subscriptions increased. Their payments were suspended for three months, which effectively amounts to a one-off 25% reduction.

An Associate Member (Founding Member) tier has been added to the subscriptions.

Free subscriber

• Notifications of new articles and products (no more than one a week)

• Access to free subscriber articles and online resources

• Seven Secrets of Successful Learning - a Welcome Gift.

Learning Pages Member

$8 per month
$80 per year ($6.67 a month, 17% saving)
These prices will be converted to your currency

• Exclusive articles

• Exclusive online resources

• Exclusive icebreaker and quotation collections

• Downloads of published PDF articles, books and programs

• Notifications of new articles and products (no more than one a week)

Associate Member

$100 per year

• Support the Learning Pages project

• Help shape what I create by suggesting topics that interest you

• Advice on learning, writing, computing and programming

• Earn by writing for Learning Pages

If a reader subscribes and takes out a paid subscription, using your article’s Subscribe button, you’ll earn 50% of their subscription fee for the duration of their paid subscription.

My share is what’s left of the remaining 50% after payment fees and Substack’s 10% fee have been deducted.

I thought that there was only one kind of reasoning, and that was logical reasoning. However, my understanding of reasoning changed when I read Being You: A New Science of Consciousness by Anil Seth (ISBN 978 0 571 3372 9).

In his book, Anil Seth describes three different types of reasoning:

• Deductive

• Inductive

• Abductive (Bayesian)

Deductive Reasoning

Deductive reasoning is one of the most common forms of reasoning. It is a logical process in which a conclusion is guaranteed to be true if the premises are true. It moves from general principles to a specific conclusion.

Deductive reasoning is used in maths, science, programming, formal arguments and philosophy.

Example

If Venus is closer to the Sun than the Earth, and the Earth is closer to the Sun than Mars, then Venus is closer to the Sun than Mars.

Inductive Reasoning

Inductive reasoning is a type of logical thinking that involves making generalisations based on specific observations or examples. Unlike deductive reasoning, inductive reasoning does not guarantee the conclusion, but suggests that the conclusion is probably true.

Inductive reasoning is used in:

• Science: forming hypotheses based on experiments

• Everyday life: “My last three trains were late → trains are often late.”

• Journalism and history: spotting trends from evidence.

Example

The sun has risen in the east every day of my life.
Therefore, the sun always rises in the east.

The conclusion is highly likely, but not guaranteed. It’s based on repeated observations, not on strict logical proof.

Abductive or Bayesian Reasoning

Abductive or Bayesian reasoning is about reasoning with probabilities. It is a form of logical inference that begins with an incomplete set of observations and looks for the most likely explanation. It’s often described as ‘inference to the best explanation’.

Like inductive reasoning, Bayesian reasoning can also come to the wrong conclusion. It is often used when speed or practicality is more important than formal proof.

Example

Observation
The grass is wet this morning.

Possibilities
It’s wet because it rained last night, someone watered the lawn, or there was dew.

Conclusion
It probably rained last night.

You’re picking the most plausible explanation, even though it’s not the only possible one.

Bayesian reasoning is used in:

• Medicine: A doctor sees symptoms and chooses the most likely diagnosis.

• Crime investigation: A detective infers the most likely cause of a crime.

• Science: Hypothesis formation based on puzzling data.

Reverend Thomas Bayes

The Reverend Thomas Bayes (1702-1761) was a Presbyterian minister, philosopher, and statistician who lived much of his life in Tunbridge Wells, southern England, but never published the theorem that immortalised his name.

His ‘Essay Towards Solving a Problem in the Doctrine of Chances’ was presented to the Royal Society in London two years after his death by fellow preacher-philosopher Richard Price, and much of the mathematics was done later by the French mathematician Pierre-Simon Laplace.

But it is Bayes whose name is forever tied to a way of reasoning called ‘inference to the best explanation’, the insights from which are central to understanding how conscious perceptions are built from brain-based best guesses.

Conclusion

Deductive, inductive, and abductive (Bayesian) reasoning each offer a distinct way of making sense of the world. Deduction gives us certainty when our premises are sound, induction allows us to learn from patterns and experience, and abductive reasoning helps us act intelligently in the face of uncertainty by weighing the most plausible explanations.

In practice, effective thinking rarely relies on just one of these forms. We move fluidly between them: forming hypotheses abductively, testing them inductively, and applying conclusions deductively. Understanding the strengths and limits of each type of reasoning not only sharpens our judgment but also makes us more aware of how we think—and why some conclusions feel compelling while others deserve caution.

By recognising when to seek certainty, when to accept probability, and when to infer the best explanation, we can reason more clearly, make better decisions, and respond more thoughtfully to the complexity of real-world problems.

Filtering systems are increasingly attempting to block spam from customers’ inboxes. Unfortunately, these filters are often quite coarse and can filter the emails customers want to receive.

Read more

I have added another membership tier - Associate Member. It is equivalent to what was previously called Gold Membership on the Learning Pages site.

People who originally signed up for Gold Membership will automatically have the Associate Member benefits.

The benefits of being an Associate Member are:

• Support the Learning Pages project

• Help shape what I create by suggesting topics that interest you

• Advice on learning, writing, computing and programming

• Earn by writing for Learning Pages

If your reader subscribes and takes out a paid subscription, using your article’s Subscribe button, you’ll earn 50% of their subscription fee for the duration of their paid subscription.

My share is what’s left of the remaining 50% after payment fees and Substack’s 10% fee have been deducted.

Upgrade

December Almanac

Interesting and notable dates for December 2025. Please leave a comment if you feel that other events should be added to the almanac.

View Almanac

Latest Quotes

November’s quotes were about:

• Simplicity,

• Problem-solving,

• Simple pleasures and

• Innovation.

Are You a Risk-Taker or Risk-Avoider?

Everything we do involves some degree of risk: starting something new, speaking up or making a change. Yet how we feel about risk varies greatly from person to person. Some of us are energised by uncertainty, eager to leap into the unknown. Others feel more at ease with stability, preferring to build on what’s already secure and familiar.

You can discover your risk personality in minutes by completing a simple questionnaire available to all Learning Pages subscribers.

Read article

Writing Behavioural Objectives 🔑

An objective is a statement of what the students should be able to do once they have completed a section of the course. It is an indicator of what learning has taken place. Learning goes on inside a student’s head and, therefore, cannot be directly measured…

Read article

The Best Way to Launch a Python App on a Mac 🔑

The simplest way to launch a Python app or script on a Mac is to double-click the script. This starts the Python Launcher, which will launch the app for you.

This works well using simple scripts, but I’ve found that when using more complex applications (e.g., those that use Tkinter GUIs), the app becomes sluggish and the layout changes.

Read article

Stock Photos 🔑

I have just added a Stock Photos section to the Learning Pages Digital Vault. There are only a few photos at the moment, but I’ll increase the number over the coming weeks.

These 1400 x 788 photos are suitable for all online uses. Paid subscribers are licensed to use these photos wherever they like, provided they have not been edited and have the following acknowledgement:

Photo by Learning Pages Graphics

Learning to Learn eBook

I have published a ‘Learning to Learn’ eBook to accompany the ‘Learning to Learn’ eCourse.

Paid subscribers can download this book from the Books section of the Learning Pages Digital Library.

You can also buy a copy from the Learning Pages Gumroad Store.

🔑 Paid subscriber access

Upgrade

“Study the past, if you would divine the future.”
- Confucius (551-479 B.C.)

Paid subscribers have access to the Learning Pages collection of over 500 quotations.

Subscribe now

Check out today’s Quote for the Day.

Confucius

Confucius was a Chinese philosopher and teacher who lived from 551 to 479 BC during the Spring and Autumn periods of Chinese history. His real name was Kong Fuzi, or Kongzi in Chinese.

He was born in Lu (present-day Shandong province) into a family of modest means. His father died when he was young, and his mother raised him.

Confucius held several minor government positions before becoming a teacher and advisor to various rulers. He advocated his ideas on governance, morality and social relationships to these rulers.

Phone-friendly version

* See The Moon and its Phases

Please leave a comment if you think of other events that should be included in the December almanac.

“Be not the first by whom the new are tried, nor yet the last to lay old aside.”
- Alexander Pope (1688-1744)

Paid subscribers have access to the Learning Pages collection of over 500 quotations.

Subscribe now

Check out today’s Quote for the Day.

Alexander Pope

Alexander Pope (21 May 1688 - 30 May 1744) was an English poet, translator, and satirist of the Enlightenment era. He was considered to be one of the most prominent English poets of the early 18th century. An exponent of Augustan literature, Pope is best known for his satirical and discursive poetry, including The Rape of the Lock, The Dunciad, An Essay on Criticism and for his translations of Homer.

The simplest way to launch a Python app or script on a Mac is to double-click the script. This starts the Python Launcher, which will launch the app for you.

This works well when using simple scripts, but I’ve found that when using more complex applications (e.g., those that use Tkinter GUIs), the app becomes sluggish and the layout changes.

I am going to describe three other methods, but before we start, it would be a good idea to check the Python version you have installed on your Mac:

Open the Terminal by pressing ⌘ + Space, typing ‘Terminal’ and pressing Return, or going to ‘Applications > Utilities > Terminal’.

Enter the following:

python3 --version

If you see something like:

Python 3.14.0

You’re ready to continue.

If not, install Python from python.org/downloads

I will also provide a download link to my ‘Jargon - Business Jargon Generator’ app that you can use to practice and learn about launching Python programs.

Read more

“I adore simple pleasures. They are the last refuge of the complex.”
- Oscar Wilde (1854-1900)

Paid subscribers have access to the Learning Pages collection of over 500 quotations.

Subscribe now

Check out today’s Quote for the Day.

Oscar Wilde

Oscar Fingal O’Flahertie Wilde (16 October 1854 - 30 November 1900) was an Irish writer and poet. He was among London’s most popular playwrights in the early 1890s. Today, he is remembered for his epigrams, plays, and the circumstances of his imprisonment, which ultimately led to his early death.