For my blog Google Analytics has gone weird

I occasionally look at Google Analytics for my blog in the slim hope it might tell me something about the level and nature of interest (or lack thereof) from my audience.

Until a few months ago, I thought the data was somewhat meaningful and believable. It gives data on the number of views for the whole blog and for individual posts. It also provides information for specific countries.

Until a few months ago, there were (supposedly) about one thousand page views per day. A typical post would attract a few hundred views within a week of posting and sometimes up to a few thousand after a few years.

However, the graph below shows how daily traffic increased dramatically at the beginning of August, rising to several hundred thousand views per day!

[Apologies for the poor quality of the screenshot... click on it for higher clarity]

Of course, at first I thought that this was exciting and flattering. However, I quickly realised this was ridiculous. This can be seen by looking at the breakdown by country.

Of the 10 million views from the past 3 months, 1.1 million came from Brazil, and about 300 K each from Ukraine, Bangladesh, Saudi Arabia, and Venezuela. Somehow, I don't think there are that many people in those countries interested in my blog!

Perhaps more importantly, if the traffic was real, it should translate into an increased volume of blog comments, personal emails, speaking invitations, enquiries from potential students... I have seen none of that.

The traffic must be driven by some sort of bots, troll farms, or AI site scraping. I asked Google Gemini whether AI could have increased my traffic. It responded

your traffic surge is a direct result of leveraging the content creation boom and capturing high-intent organic traffic ahead of your competitors [1].When ChatGPT launched, it created an immediate divide between blogs that adapted quickly and those that ignored the technology.

This is wrong, as I completely ignored the technology! I have never used it to write content. 

It also provided me with an unrequested "Step-by-step plan to protect my gains." [Shameless AI self-promotion].

A bit more data. The number of views of individual posts has not changed dramatically and still seems reasonable. Overall traffic on my other blog has also increased, but not in such ridiculous and unbelievable ways.

Overall, this now makes me sceptical about whether any of the Google Analytics data is meaningful. It also makes me more doubtful about the massive traffic claimed by some prominent bloggers.

This is all weird. Do you know of other bloggers with similar experience?

What is your experience of using AI for research in condensed matter theory?

 I have been dabbling a little with using AI (at a very basic level) to help me with some research problems. For example, in a recent preprint, I did the maths in the Appendix with help from AI.

I am curious to hear from the experience of others of using it for research in condensed matter theory and theoretical chemistry. 

Peter Woit made a similar request to the high-energy theory community. The responses are helpful and worth reading. An earlier post about AI in math drew the following concrete suggestions from "Greg"

"In math, one thing I found useful to get my bearings was to go through some of my published papers and ask it to try and improve the main result.

Another is to do an adversarial audit of a paper you are very familiar with (yours if you have the stomach for it, someone else’s if not) and then see which of the complaints are pedantic and which are actual.

A third is to give you a history and references/citations for an argument that it provides.

A fourth is to suggest different ways to simplify an argument, even if they are only half-formed in your mind or if you are having trouble making them precise. (Try to be as precise as possible, but don’t use the same filter that you would use with a colleague. Sometimes vague is the best you can do.)

If you are thinking of proposing a question or conjecture in a grant application or preprint, ask it to stress test the conjecture. Don’t lean too much in one direction when prompting it and asking repeated questions (unless there is a particular thread you are trying to untangle, the use your nose, of course), but try to have it explore the positive direction in one query and the negative direction in a follow-up."

[A year ago, I posted about a benchmarking paper for condensed matter theory. I have not seen any updates on that work. For example, have the latest AI models improved significantly on these benchmarks?]

About a month ago, my office mate, Henry Nourse, gave a nice talk, "Is my skillset safe in an AI future?" at the UQ Condensed Matter group meeting. Henry gave a very concrete example of how he used it for a problem involving Raman scattering from quantum spin chains.

Henry stimulated me to see whether AI could help with a problem I was struggling with. It was basically a mathematics problem involving a Fourier transform. There were things that I thought were probably true, but after days of messing with algebra, my aging brain could not prove them. Google Gemini gave the answer in less than a minute. On the one hand, that is impressive. On the other hand, the absolute key was knowing precisely what question to ask it and knowing whether the anwer made sense. Furthermore, it took hours for me to check the result and to refine it, sometimes through further prompts. Besides getting the result, a nice benefit is that AI can quickly generate a LaTeX file that describes the mathematical argument and provides relevant references. Nevertheless, there are still hours more work of checking and polishing the text.

[Aside. The proof made use of various Bessel function identities that I had forgotten. I had used them extensively in a paper about thirty years ago.]

This success motivated me to use Google Gemini to develop, start to solve, and analyse a "simple" classical two-dimensional model of springs interacting with Ising spins. It is simple enough that I think most of the analysis can be done analytically. I have been working on related models on and off for the last couple of years. However, again due to my aging brain, I often get stuck on algebra or the subtleties of the theory of elasticity. AI does the algebra quickly. Furthermore, all the equations are formatted in LaTex. Great!

Again, it bears repeating that a thorough knowledge and intuition about the underlying physics is probably essential for knowing what prompts to make, and whether the answers make sense and are useful. The fact that I had struggled with related models offline beforehand really guided my interaction with AI.

Again, everything still needs to be checked. Sometimes AI makes suggestions that I think may be "hallucinations" [e.g. suggesting connections to the Svetitsky-Yaffe conjecture for lattice gauge theory] that may have arisen from earlier questions about unrelated topics. When I ask similar questions on a later day, it sometimes gives a different answer. For example, once it neglected to include the effect of one term in the Hamiltonian.

Initially, I struggled to keep records and manage the workflow. Eventually, I found it helpful to have a single LaTex file which is like a rough draft of a paper, into which I insert results. I found my engagements with AI both interesting, stimulating, and stressful. I now limit my sessions to short periods of time and then polish and edit the LaTex document in between sessions. This helps me consider what the next steps are. I am glad I am retired and can do all this at a leisurely pace.

What I am doing is incredibly basic compared to what may be possible. I am just using the free version of Google Gemini.

My limited experience raises all sorts of questions about the potential and pitfalls of graduate students using AI in research. Will it enhance or hinder their education? Potentially both?

There are some great quotes from Terence Tao in a recent New York Times article

“The effort needed to solve problems [without AI] is often very instructive. It teaches you something. It’s like going to the gym and having a goal to lift a weight a hundred times. Now, A.I. can solve questions without really getting any value out of them. It’s like having machines that can lift weights for you at the gym.”

“It feels to me like a really clever student who has memorized everything for the test but doesn’t have a deep understanding of the concept,..”

“This has some value,... But once someone shows a specific path to the waterfall, people just take that path. They don’t spend as much time looking for other paths.”

I also recommend Tao's recent lecture at the International Congress of Mathematicians.

For now, I would love to hear from others, particularly in condensed matter theory or theoretical chemistry.

What do you use AI for? What have you learned from the experience? How do you go about it? For example, how do you decide what prompts to use? How do you manage workflows? How do you document your work? What is it good at? What is it bad at? What potential benefits do you see? What pitfalls?

The subtlety of elastic interactions in spin crossover materials

My collaborators and I recently posted a preprint

Exact mapping from short-ranged harmonic elastic models of spin crossover materials to Ising models with interactions at all length scales

Nadeem Natt, Gian Ruzzi, Jace Cruddas, Ross H. McKenzie, Ben J. Powell

Spin crossover (SCO) materials are reversible molecular switches found in a wide range of transition metal complexes and metal organic frameworks (MOFs). They exhibit diverse spin state orderings and transitions between them. Here we present an exact mapping from harmonic elastic models to Ising-like models with both a short-range Ising interaction that decays with a power law at large distances and a long-range (infinite-range) Husimi-Temperley interaction that is independent of distance. We apply this mapping to a simple model of SCO frameworks. This provides a microscopic justification for an Ising-Husimi-Temperley model description, which has previously only been justified on phenomenological grounds. Elastic frustration is required for non-zero Ising interactions, but whether or not the short-range interactions in the Ising model are geometrically frustrated depends on the ratio of the bulk and shear moduli, or equivalently Poisson's ratio. The long-range interaction has two origins: (i) a self-interaction on the average spin state, mediated through the coupling between the average spin state and the unit cell parameters; and (ii) an infrared divergence in the spin state-spin state coupling mediated by displacements of metals and ligands within the unit cell. In the absence of elastic frustration these terms are equal and opposite so there is no long-range interaction. However, in general they do not cancel and there is a long-range Ising interaction. The long-range interaction is independent of the distance between metal centers, nevertheless it leads to an extensive contribution to the (free) energy. In this model the Husimi-Temperley interaction dominates transitions of spin states, whereas multistep transitions and intermediate order are observed if only the pure (power-law) Ising interaction are retained, only single-step transitions are found in the full model.

The paper has a long history. The original version was posted six years ago! 

Our earlier derivation of the effective Ising model was incorrect as it did not take into account subtle boundary effects that lead to an infinite-range Ising interaction. This significantly changes the phase diagram of the model. For example, multi-step spin state transitions do not occur. 

It required a new generation of graduate students to redo the calculations and finish the paper. 

The paper also includes a first for me. I did the derivation in the Appendix with AI (and checked by a human). This experience showed me both the potential and pitfalls of AI in my research. More on that later.

What does this movie tell us about the modern university?

Last night, my wife and I watched the movie, Wit.

You can watch the full movie here (free with ads). I should warn that some of the content is quite disturbing and confronting.

I asked Google Gemini the following question:

"the movie wit, starring emma thompson, features two professors, one of oncology and another of english literature. the latter has terminal cancer. in spite of their academic credentials, neither seems equipped to cope with the human side of the suffering of the patient. alisdair macintyre was a critic of the modern university. what might have he said about this movie?"

Here is the answer.

"Alasdair MacIntyre, a prominent critic of the modern university and contemporary moral culture, viewed modern institutions not as communities dedicated to a shared human good, but as bureaucratic structures driven by specialization, technical expertise, and managerial effectiveness.

Based on his philosophical framework—most notably developed in After Virtue—MacIntyre would likely view the dynamic in the movie Wit as a quintessential illustration of the moral and intellectual failures of the modern university. Here is what he might say about it:

1. The Triumph of Bureaucratic Expertise Over Human "Telos"

MacIntyre argued that modern academic disciplines have abandoned any coherent notion of a unified human telos (purpose or ultimate goal). Instead, they have fractured into compartmentalized technical specialties.

• The Oncologist: Represents the epitome of modern "technocratic" or "managerial" expertise. To the hospital and the university, medicine is treated as a problem-solving science rather than a moral practice aimed at human flourishing. The patient is reduced to a data point, a mechanical puzzle, or a vehicle for research.
  
  
• The English Literature Professor: Though her field deals with the humanities, the modern study of literature has similarly grown hyper-specialized, detached, and academicized (focused on obscure analysis, such as John Donne’s holy sonnets, rather than wisdom or existential truth).
  
  

2. The Failure of Emotivism and Detached Rationalism

MacIntyre famously claimed that modern moral discourse has devolved into emotivism—the idea that ethical judgments are merely expressions of personal preference or feeling rather than rational truths.

• Neither professor is equipped to handle human suffering because their education has stripped them of a shared moral language.
  
  
• The literature professor spent her life analyzing deep themes of mortality and grace on paper, but because modern academia treats texts as objects of clinical critique rather than guides for living, she remains entirely unprepared to face her own mortality with actual grace. Her knowledge was purely aesthetic and intellectual, leaving a void where real moral virtue should be.
  
  

3. The Loss of Genuine Community

For MacIntyre, universities were historically meant to be traditions-based communities focused on cultivating virtues.  He viewed the modern university as having traded this for bureaucratic careerism and bureaucratic compartmentalization.

• In Wit, the coldness of the hospital-university complex highlights a complete absence of genuine human community. Interactions are transactional and analytical.
  
  

In short, MacIntyre would likely argue that Wit demonstrates the tragic irony of the modern academy: it produces hyper-competent specialists who know an immense amount about narrow subjects, yet are entirely illiterate in the art of being human."

That is a beautiful reflection. My enjoyment of it is somewhat overshadowed by my awareness that AI produced it. It also raises all questions about how to use AI in ways that do not diminish my own humanity.

Questions to consider when evaluating AI

 I am a slow adopter of new technologies. I have recently been playing around with AI at a very basic level on some research problems. Colleagues are also telling me about their experiences. I want to write something about its potential, both good and bad, for research and teaching. I want to hear from more people, particularly in condensed matter theory. However, before that, I think it is worth stepping back and asking some bigger questions than "Can AI help me publish more papers?" or "How do we stop students cheating on assessment?", as important as they are. I know the mathematics community is going through some angst and has issued a declaration about AI, and it is good to see that level of reflection.

Neil Postman (1931-2003) was a media theorist and cultural critic at New York University who spent a lifetime wrestling with questions about the broader implications of new technologies. In a talk given in 1998, he considered five things we need to know about technological change. Postman's enduring influence and relevance are marked by the fact that these five things featured in a column in The Washington Post, "Is the Internet Evil?" by Christine Emba, published in 2018.

Below, I summarise the five ideas from Postman's talk and provide questions (in italics) we should ask about any technology, particularly Artificial Intelligence (AI).

1. All technological change is a trade-off. 

"the greater the wonders of a technology, the greater will be its negative consequences" 

Don't just ask the question "What will a new technology do?" Also ask, "What will a new technology undo?"

"a sophisticated perspective on technological change includes one’s being skeptical of Utopian and Messianic visions drawn by those who have no sense of history or of the precarious balances on which culture depends."

Through adoption of the technology, what will we lose, individually and as a society?

2. The advantages and disadvantages of a new technology are never distributed evenly among the population.

Who will benefit? Who will be harmed? 

Winners will try to persuade losers that they will benefit as well.

Benefits and harms can relate to employment, finances, social status, health, and political power.

3.  Embedded in every technology are powerful ideas. 

"These ideas are often hidden from our view because they are of a somewhat abstract nature. But this should not be taken to mean that they do not have practical consequences."

"The telegraphic person values speed, not introspection. The television person values immediacy, not history... the computer person values information, not knowledge, certainly not wisdom." 

"The medium is the message."

What ideas are embedded in the technology?

How does it make us use our minds and bodies?

How does it affect our personal relationships and social cohesion?

4. Technological change is not additive; it is ecological. It changes everything.

"The consequences of technological change are always vast, often unpredictable and largely irreversible."

The entrepreneurs who started the television industry "did not mean to turn political discourse into a form of entertainment."

The consequences can be social, economic, political, environmental, religious, and health-related.

What are the unintended consequences of the technology?

5. When a technology becomes mythic, it is always dangerous because it is then accepted as it is, and is therefore not easily susceptible to modification or control.

"...our enthusiasm for technology can turn into a form of idolatry and our belief in its beneficence can be a false absolute. The best way to view technology is as a strange intruder, to remember that technology is not part of God’s plan but a product of human creativity and hubris, and that its capacity for good or evil rests entirely on human awareness of what it does for us and to us."

How does the technology lead to idolatry? Do some people worship it, its creators, or its owners?

Finally,

Do we use the technology or does the technology use us? 

In different words, will we shape our lives to fit the requirements of the technology, rather than have our values shape our use of the technology?