Using Machine Learning for Writing IT Media Articles?

Executive Summary

  • Why many articles by the IT media have such a low-quality level they could be written by AI.


Recently I was reading an article about how SAP is dropping support for SAP Mobile. The article that covered this topic was actually the following text.

“SAP has decided to kill off its Mobile Platform.

An announcement from January 31st pointed out that support ends for the SAP Mobile Platform on December 31st, 2020, but added news that “we do not plan to offer Extended Maintenance.”

SAP will offer “customer-specific maintenance”, an ongoing paid service as the default option for customers. The service offers “customer-specific problem resolution” but no new code and “no legal changes”.

In a colossal non-surprise, the post also mentions SAP’s Cloud Extension model as a fine alternative. Indeed, the company kindly gave its users three reasons to consider it: lower costs, the flow of new features and integration with the rest of SAP.

Paying to keep an application on life support versus adopting a product your vendor actually cares about isn’t much of a choice.

With three years to decide on one option or the other, The Register imagines SAP Mobile users’ decisions won’t be hard to predict, despite the company’s wares often being so fundamental to business that changes are contemplated with trepidation..”

That is the entirety of the article. It is 176 words long!

IT Media Entities Offering Zero Analysis

Does the Register not know that SAP Mobile has been dead for years? (they state it will surprise no one, but that is not a sufficient explanation.)

This announcement will affect no one because there is no one to support for SAP Mobile.

This would be like me announcing Brightwork Research & Analysis is dropping all support for the exhaust systems Lamborghinis.

Except the things is, we don’t do automotive repair at all.

IT media essentially relies on the fact that people are generally not aware how the system works. But the more one analyzes the system of IT media, the more it becomes apparent that the entities are tricking their readers. It takes a while to figure out what is going on, but once you do, articles from most of the IT media entities follow predictable patterns. One thing is very clear — the IT media entities do not care either what is true, or in contesting any of the ideas and information presented by the largest entities in IT. 

Why Our Esteem for IT Media Keeps Dropping

Its very much about copy and paste, and get paid.

  1. Go get a quote
  2. …include that quote
  3. …and move on.

The idea that if you gather different quotes then you are doing journalism is ridiculous. In the SAP space, the quotes in these types of articles are almost exclusively from entities with a financial bias. Either they are from SAP itself, an SAP consulting firm, ASUG, which is a user group that is controlled by SAP, Gartner, which receives massive amounts of money from SAP. None of these biases are explained to the reader, so the sources are simply presented as individuals and entities with information to provide. Very few alternative viewpoints are offered.

And if SAP is your advertiser, why would we expect any?

The Register not interested in explaining history to their readers. In the article on SAP Mobility, did The Register go back and check SAP’s statements about how they were going to take over Mobile? They did not. But we did this article How Accurate Was SAP on Mobility Due to the Sybase Acquisition?

It turns out that SAP was incredibly inaccurate in its statements about Mobility. Why is that not relevant to explain to readers?

If an entity is highly inaccurate about previous statements, why would anyone listen to them about current statements about the future? If a media entity simply repeats statements from entities that pay it without validating the source, what is the purpose of that media entity except to deceive the reader?

It almost appears that SAP controls the volume nob of IT media. When SAP has something new to sell, and inaccurate claims to make, IT media entities are there to write long high octane articles that SAP wants. But when something is discontinued, IT media reliably turns down the volume, writing the shortest article possible. This was covered in our article How SAP Has Quietly Changed Strategy on HANA and Oracle. This was a change in policy regarding Oracle and extending their relationship. SAP did not want this change in policy known. The article we found on this was again……curiously short at 144 words. 

Lets put this in the proper context.

When SAP announces something, it gets all the benefit of the doubt, but when something from SAP dies, (such as the previously overhyped SAP Mobility) it dies quietly.

No analysis, no explanation, nothing.

This particular article about SAP Mobility is nice and short, exactly as SAP wanted it. SAP does not want The Register pointing out what SAP stated in the past about SAP Mobile, its embarrassing. Therefore, the article is written to observe the fact, but to make the article as innocuous as possible. 

The Register, ComputerWeekly, Forbes, etc.. are all there to make their customer….look as good as possible.

Introducing The IT Media Automation Article Algorithm (ITMAAA)!

We have been investigating machine learning for forecasting. ML has a lot of marketing puffery attached to it. And you don’t need to run an ML algorithm to simply automate tasks. The problem of forecasting is much more complex than IT media article writing

Let’s look at the algorithm for publishing on an IT subject.

  • Step 1: Find IT term with a high number of searches.
  • Step 2: Maintain a table of vendor contacts.
  • Step 3: Maintain a table of vendors that say they do different “keywords.”
  • Step 4: Maintain a table of desirable outcomes that could be attributed to anything such as “digital transformation,” “improvement initiative,” “Six Sigma program,” etc…
  • Step 5: Send out automated emails to vendors that are listed in the DB as doing that thing.
  • Step 6: Request several quotations from the vendor. Further quotes are provided by references to either end customers or consulting companies that are provided by the software vendor in the file response.
  • Step 7: Send automatic request for money to be included in the article.
  • Step 8: Include quotations from vendors that meet the minimum threshold (say $7000).
  • Step 9: (IMPORTANT!) Provide no analysis or fact-checking of the quotation.

With the input material available, now we move to applying the writing algorithm which inserts quotations into a sequence on the keyword.

  1. Introduction topic is important because <Keyword> is enabling companies to <Desirable Outcome>.
  2. Quotation #1
  3. Glue Paragraphs:(These can be written by inserting sections talking about how “keyword” is interesting and leads to “insert generic keyword like — “digital transformation,” “improvement initiative,” “accelerated business value,” etc.. List of glue paragraph keywords kept in database that is used by all articles.)
  4. Quotation #2
  5. Glue Paragraph
  6. Quotation #3
  7. Glue Paragraph
  8. Quotation #4
  9. Conclusion

That is the algorithm. Now here is the machine learning part.

Each article is tracked for how much income it brings in. As time passes the “machines learns” which articles can pull the most money out of sofware vendors and software consulting companies. For instance, articles eventually will look something like his.

  • “SAP HANA Cures Cancer”
  • “Larry Ellison Voted the Best Looking Man in the Universe”
  • “Deloitte and Accenture Delivery Unlimited Value, Resulting in a 6 Month Backlog in Beginning New Projects”

Why Have Human Involvement Performing Writing Side of IT Media Entities?

Writers take up time and resources that could be better spent in the employment of the primary purpose of the media entity, which is to obtain funding from industry sources. In fact, there are many articles which are written by the industry source itself. Not having to produce any content (at least with a human) is the ultimate objective of IT media entities. That is we have IT media entities that would prefer to do no work at all. They merely desire to rent out the access to its audience to a funding source. These paid placements in most articles are not identified as paid placements. They can be identified by several techniques, including how well the text matches the marketing material from the software entity as well as the links that exist on the page.

But the funders can’t be expected to write all of the content for the media entity unless the entity is like DigitalistMag, which is in fact owned by SAP.

Having to write articles is where the ITMAAA comes in.

Creating ML algorithms to say produce a better forecast is complex, but automating media is far easier. The beauty is that IT media has already moved in that direction by making its writers produce puff pieces, shrinking deadlines, reducing pay, having writers who don’t know the subject areas they are covering.

Complete machine article development is just the next step.

And the thing is that this algorithm and minimally written articles come from entities that dominate the IT media sphere. The Register gets 11.4 million page views per month! The less they ruffle feathers at their funders, and the more they serve disposable and easily digestible articles, the more money they make. Like IDG. They claim to be #1 in IT media. Yes, #1 in selling out that is.

Keeping the Accuracy Deliberately Low

We review many articles from IT media for accuracy. IT media entities like to keep their accuracy below 5 out of 10. Any higher takes doing research and runs the risk of offending the funders.

The Importance of Profit Maximization in Media

If we look at the top media entities, like IDG or Forbes or TechTarget, they are all run simply to maximize profits and have no concern about publishing accurate information or even actually being journalists. At one time I thought that IDC/IDG was a legitimate entity. But they aren’t. They are media entity that publishes anything for money. They own 8 of the top 20 most visited IT media websites, including CIO Magazine and ComputerWorld.

IDG’s website is quite clear that they provide passive buyers ready to make a purchase. What types of articles does one think they might end up publishing with this is a primary objective? 

Quality Publishing from….China?

A Chinese conglomerate owns IDG. A Chinese conglomerate owns Forbes. China is a country which is firmly opposed to the freedom of speech in all of its forms. In China, if you were to be so “out of it” to publish anything against the government, you would find yourself in prison exceptionally quickly. You may or may not exit that prison. And if you do, you may not look or feel very much like the same person as when you went in. That is who western readers should be getting their information from? From a conglomerate based in a country that has both never discovered and does not respect either the freedom of speech or the ability to question authority?

And there will likely be more Chinese purchases of media entities in the future.


IT media is now predominantly advertising arms available to the highest bidder. These entities pretend to run journalistic outfits when in fact they don’t care about reporting anything but what they are paid to report.

However, it’s time to automate this process. Money is being left on the table by having articles with human involvement. This is not profit maximizing and could lead to shareholder revolts in these companies.

To satisfy shareholders, most of the employment in IT media needs to be allocated to the sales arms that work out financial terms with customers of the IT media entities, which are the corporate placers of advertisements. In the future, the only real jobs in these media entities will be in ad sales.