As will become apparent from posts I will be making here over the next few days, a lot of agonising and thought has gone into the future direction of this blog and the other channels we have over the last few weeks.

That process, which has been heavily influenced by the involvement of Thomas and James in the production processes of my work, has looked at questions around:

• what we are doing,
• why we are doing it,
• how we best communicate,
• how we reach different audiences,
• how we make the work more accessible, and
• how we fund this work on a resilient basis.

This has challenged all of us to come up with new answers, having recognised that not everything we are doing is working to the best effect.

Having reviewed some of these issues, we also identified some specific current weaknesses.

That process started with a recognition that our podcasts were undermining our channel's reach on the YouTube algorithm, which is why we have not done any for the last few weeks. The likely response is that such content will move to Substack soon, on which subject I will have more to say soon. John Christensen and I are discussing making a series on capitalism for that channel.

We then recognised that there are real issues with this blog and its current structure that need to be addressed, which will probably require, in the end, a complete redesign, as this version of its design is now nine years old.

The search problem

A particular problem that we found with the blog is something of which we have been aware for a while, but which has challenged James as he takes on a more editorial role within the work that we do, which is in finding just what I have written on a subject before now, and what the key messages on any topic that I have discussed might be.

Our existing search engine is based on search techniques that almost nobody understands or uses these days, and the consequence has been that the content of more than 25,000 blog posts has become largely inaccessible. This, as he has pointed out with some frustration, has made it a seriously underexploited resource.

As a result, we explored improving the blog's search facility, and there are AI-powered search engines available for this purpose, which would be considerably easier to use than the one we currently have, but they are also expensive. The likely cost of integrating them would, in search cost terms, come to potentially thousands of pounds a year.

That then linked straight into our funding question because we are funding the whole of this operation on a financially marginal basis at present. That means that it does cover its costs right now, but only on the basis that any surplus it happens to make represents a rate of return to me on expenditure of effort of considerably less than the statutory minimum rate of pay for the time involved.

Admittedly, I now have a pension, which means I am not looking for a significant financial return from this activity, but equally, I cannot afford open-ended risk, which such a search facility could create, so we looked for another way to access the archive.

This is a slightly long way of introducing another post that I will be publishing this morning, which is entitled ‘Richard Murphy's View on Modern Monetary Theory'.

This has been written in the same AI-assisted way that the AI-assisted Economics Questions series was generated, although with the twist that all the source material used was entirely drawn from this blog by treating it as a large language database in its own right.

The resulting post is written in the third person, analysing my work from that perspective, as this piece, and others in a series that might follow if this one is thought to be of use, are intended for sharing and outreach, as well as explanation and, to some degree, as a replacement for a search facility.

The prompt that generates the post is absurdly long, and the output is, of course, subject to editing and checking. These are processes in which Jacqueline and James are involved before I give it a final once-over. We are using Claude AI for this purpose. Although the prompt was initially developed in ChatGPT, it turns out to work much better on Claude.

What happens next

Our aim is that all of these posts will be available in due course as formatted PDF files. When this blog is redesigned, they might get an organised home here. In the meantime, they might get a home on our Substack site, which James is now developing. If it ends up looking like Robert Reich's, so be it.

The question is, is this post useful? Would a series of such posts be useful? So far, I have about 90 draft titles. I am open to more. And would PDF downloads from a neatly organised archive also help? Comments would be welcome. So would answers to this poll, and thanks in advance.

 Loading ...

PDF of article