Back in the early days of the Internet, Paul Graham, then working for Yahoo, wrote a now-famous article, “What the Bubble Got Right” (2004). Graham’s article was written shortly after the dot.com bubble, the sudden escalation in the valuation of online companies, as can be seen from the NASDAQ index spike around the year 2000. After that bubble subsided, many believed that things would return to the pre-bubble state, but Graham identified several by-products from the dot.com boom that seemed to have persisted beyond those crazy few years. What does Graham list?
- Retail Venture Capital
- The Internet
These are all things that have remained with us since the bubble – and most of them are still with us today. Graham’s post is well worth reading, if only for the sense of optimism that surrounded the Internet culture in the first few years.
John Thornhill in the FT (22 September 2022) considers Graham’s list, and suggests some present-day equivalents. Unfortunately, his list (comprising only five topics) is rather underwhelming, and, I believe, rather short-term. They comprise:
- The importance of data
- Decentralized workforce
- Energy transition creating stock market winners
- Trading digital assets (using Blockchain)
I’m certainly in agreement on the importance of data, but I’m not sure that the last three of these are really the most important present-day implications of the Internet. I’m sure we can do better than that. Here is my list of ten present-day benefits of the Internet (in its broadest sense, including but not limited to the new dot.com boom and bust companies):
1. Digital Information
The entire knowledge process moving to digital is a transformation with profound implications. It has by no means fulfilled its promise. I remember being underwhelmed when I saw some of the first digitised content from Oxford University Press and Microsoft (anyone remember Microsoft Bookshelf?) It was simply the same words, on a screen. That took just a few years. Learning to make effective use of digital content will take much longer.
2. The democratization of knowledge
This is quite simply a fundamental shift in attitudes. The move towards open-access publishing for academic content is simply one manifestation of this. Access to books is just one aspect: The Internet Archive, making the world’s content available in digital form, is another. A single collection available on the Internet Archive, The American Library collection, comprises some 3.4 million books, most of them available for free loan. It is becoming possible to study without access to the many subscription-based institutions, such as academic libraries. Making more books freely available on the Internet might not increase the general level of intelligence, but we can all agree the principle of the public library is a valid one.
Using such knowledge, questions that used to be deferred to an authority are now dealt with via the Web, perhaps not instead of, but certainly as well as, consulting an expert: think medical diagnosis, or legal opinion. Anyone visiting their local GP will no doubt look up their symptoms, or condition, on the Web. And let’s remember Wikipedia here: a model of a collective resource providing free content, where formerly you had to pay; and created by volunteers.
Suddenly we had more choice than ever before. In the High Street you could buy perhaps two or three types of anything. With online trading, there were hundreds of things available, and hundreds of suppliers; so many that it rapidly became impossible to see all the options.
Of course, we slowly began to realise that many buying choices are delivered by just one supplier: Amazon (see (6) below).
4. The Recommender
What! Such a trivial tool as the recommender? Well, as a direct result of a proliferation of choice (3), we were all confronted by a bigger choice than we could manage. The online age stepped in and provided us with a new tool: the recommender. Suddenly, sites proliferated offering the top ten of this, the top ten of that. The site Five Books is quite an interesting example; what is interesting is that it could not have existed in the 1960s. When you think that Spotify and Netflix are essentially recommender tools, and Google is one vast recommender service, you start to recognise the impact recommenders have on today’s world.
5. We all think like startups today
Startup mentality, and the consequent awareness of capitalism: the cycle of boom and bust, growth and failure.
In those twenty or so years since the first digital bubble, we have all lived the startup dream. We have all read about startups, and pivoting, and minimum viable product, and crossing the chasm – even if, I guess, we worked for Shell all our working life. So that is one change: we all approach things as if we were a startup. I believe there is a much sharper awareness of the business potential of new ideas today; the downside is an endless obsession with monetization. The promise of getting rich quick dates back, of course, to the South Sea Bubble, and probably earlier, but the dream of instant wealth is, I think, more powerful than ever following the spectacular rise of a handful of internet-based companies.
6. Big is better … is it?
When I search for stuff to buy, it is undoubtedly easier to search on one site: Amazon. When I want to listen to music, I look it up on Spotify. If I want to see a video, I start with YouTube. If I want to know where something is, I look it up on Google Maps.
The five FAANG companies (Facebook, Amazon, Apple, Netflix and Google), particularly Amazon and Google, tend inevitably towards a monopoly – and that’s good news, initially. The provision of search tends towards a monopoly, in the sense that for consumers, it makes sense to query the biggest corpus. Similarly, it makes sense to use Amazon if it provides the biggest range of products than other retailers. However, monopolies turn out to have drawbacks – as users of Stagecoach buses in the UK soon realise.
7. Data analysis, and the new importance of statistics
As John Thornhill points out, data is vital, but specifically, the advent of digital processing means that we can now track activity. What did users click? What did this process involve? Data analysis enables more precise interpretation of myriad operations, from article citations to sports players (“the midfielder had the most touches (67), attempted the most passes (45) and completed the most accurate passes (41) by an England player”).
8. Digital workflow
Another long-term development is the steady rethinking of all business workflow by digital processes. Does this process require human intervention? Can we substitute a machine for part or all of it?
9. Social media
Look at any public space and you can see people gazing down at their phones. Social media must be the most significant change of the last twenty years. We have immediate contact with our friends and relatives. We can tell our loved ones “I’m on the way” every two minutes. We can share experiences and special moments.
At the same time, social media has turned us all into public showcases. It is now difficult to live without telling others what you are doing. There are complaints in newspapers about people who don’t respond to WhatsApp messages. A Lionel Messi goal is only real, it seems, if the spectators capture it on their phones and share it with others. This is the exhibitionist age. Job CVs have moved from understatement to glorious overstatement (“self-starter”, “passionate”, “influencer”, “created”, “initiated”, “delivered”, are the terms you need to describe yourself in 2022).
Social media, of course, enables venom and trolling – but that’s a subject for another post.
10. Artificial Intelligence
The invention of XML was seen as a major step forward in its time, but interoperability turns out to be necessary rather than sufficient. Digitisation was simple, but developing true interoperability of data requires far more work. Moreover, you often don’t see good results until the metadata is complete.
Artificial intelligence, on the other hand, is here now and delivering results. I don’t mean replicating the human brain; by artificial intelligence, I mean that one small part of it, “narrow” AI, the use of statistical and Bayesian tools to take a corpus and to draw inferences from it, using probability. This switch, from precision to approximation, is after all one of the great achievements of the Google search, which delivers results from imperfect metadata. Google indexes all pages, however they are coded (with a few exceptions). Google could wait until every site it indexes has perfect metadata, but then we will never get any search results, or we can learn to make use of probabilistic results from the imperfect data provided. And we are happy, for the most part, with the imperfect but probabilistic results that Google gives us.
So there are my ten benefits of the current Internet bubble. Perhaps it is a less entertaining list than Paul Graham provided back in 2004 (I don’t think anyone today would include “California” or “nerds” as a positive effect of the Internet); but, however positive a spin you try to put on it, some of those bubble benefits turn out to have significant drawbacks. In the early years of the 21st century, Google (founded 1998) could still say “don’t be evil”, but, several years later, it turns out to be not quite so simple. The coming few years will show how we manage, or if we can manage, to deal with those shortcomings, from fake news to trolling. Is the balance of the bubble still positive? Just about.