1
One of the easiest ways to predict the future of software is to imagine another communication amplifier. Alan Kay said that many decades ago.
Indeed, so far, the most successful means of predicting of the future of software has been this conjecture, that the next big thing will be another communication amplifier.
Apparently there is plenty of room in the market for such things.
Communication is too importantly and prominently involved in too many other important and prominent human activities. People can avoid most other new products; but most people cannot avoid a new communication amplifier when it appears. Almost no other product has such properties.
Once some people use a communication amplifier many other people who don’t use it have an enormous incentive to begin to use it, even if they don’t want to use it. Some users, some competitors, some friends, will use it to significantly decrease the cost of doing most of what else they are already doing or want to do. That affects most of what else these nonusers are already doing or want to do.
Because communication obviously factors in nearly all human activities, enough people tell other people about each new communication amplifier, and using it prominently in their other activities, other people see it being used and talk about it, and simultaneously, each communication amplifier reduces the cost and increases the ease and rate and amount of human communication, first of all and especially regarding itself.
And most possible communication amplifiers must be software. They cannot really be anything else.
So every decade one or two or three new communication amplifiers appeared in the world. These products and platforms changed the world. People flocked to use them. Then they mostly continued to use them.
And then yet other people took each new communication amplifier as an example of what to do. Each successful communication amplifier created a whole new market and countless similar products and platforms and countless complementary products and platforms. People in the world became more and more connected.
Most new communication amplifiers stack with previous, earlier communication amplifiers. Enough people continue to use previous, earlier communication amplifiers. They merely use them only a little less and the new amplifiers a lot more; they can do that because, after all, the cost of communicating, what they are doing when they use communication tools, had further decreased in that case.
Communication and information will become more and more abundant and attention will become more and more scarce. Herbert Simon said that even earlier. And that too has happened. But that should surprise nobody. Considering all the amplification of communication happening.
2
Communication amplification is about lowering transaction costs.
(a) That meant the next communication amplifiers had an easier and faster way to spread and get an audience.
Deep path dependence is part of why communication amplification was and is such a big software trend.
Transaction costs went down, courtesy of various earlier and competing communication amplifiers, and already came to exist larger social groups who, if some key members in the group did switch, then much of the whole group, altogether, would switch, or, even more often, at least, start using both platforms or products. After all, the costs of using both decreased.
Comment. Reducing transaction costs has enormous consequences. Fred Hoyle had suggested that, even though productivity can increase unbounded with greater and greater population, most social structures don’t scale, and if there will be a failure of a civilization, that is where. Transaction costs are a big part of why.
(b) This more than offset in enough cases the network effect that starting a new social network when most people are already using another social network is much more difficult than starting a social network that has no competitors in an empty market.
We can see that the theory of low hanging fruits is not quite correct. It is not true that so-called low hanging fruits get picked until nothing is left in a market. That technical and product and platform opportunities become exhausted.
Yes, most communication amplifiers are platforms or products that involve network effects, which often leads to a single competitor soon getting all or almost all the users, as people will end up using mostly what their friends use, small differences in users count magnifying into enormous leads very soon. But the point is, each such communication amplifier is really a communication amplifier. It creates larger pre-existing, more strongly connected communities, and genuinely reduces transaction costs, such that starting out with the next communication amplifier is actually much easier. It becomes easier and easier to grow quickly, if any growth does occur.
The competition of social networks is greater than between other competing software systems, because every success of one amplifies the loss of the others if users are not using all of them together. And users most likely are not using all of them together. And social networks are unique in how much time they consume. They compete with every other product on the market that takes time to consume: games, films, music, and even logic, thinking, playing, listening, because human beings enjoy social status and social interaction above all and there are only so many hours in a day. Attention is capped.
Software whose value to a user depends on how many other people are using it too starts with a much lower value to users, when it is first released and there are few users, and most other users are using competing software, than its features suggest were the competition merely about features.
But the greatest cost is early growth and all the prior communication amplification, where it is genuine amplification, tends to make that easier overall. Two opposite forces result.
Not all communication amplifiers compete.
A software application that was about sharing photos more easily gains more users and this more quickly if there is already a dominant software application whereby most people use to share texts. The service provider and maker of one, later, perhaps, buys the other. Quite predictable. But this leaves open niches for sharing videos. Then for sharing videos about playing games. And then streaming games. People hear about new platforms and tools more and more rapidly. They are more interconnected. Connection costs less and less after all.
Eventually a new social network that has many features appears and is slightly better overall and some groups switch over. At which point which social network most users use for texts may suddenly change.
It depends on the balance of these opposite forces that are produced by the same sources.
How much larger are the existing social groups that formed due to lower transaction costs to communicating, forming groups, organizing, groups that may switch or spread the word about a new social network on the established social network, versus how much more value an established social network with many users has compared to a new social network without so many users.
(c) True, overall software diversity has profoundly declined, compared software diversity decades in the past.
Despite it being easier than ever to create new software with great, original features. And despite hardware becoming less and less expensive and that making so much more software feasible. Seymour Papert, back in 1968, said that the primary limitation in getting most ideas to work was the fact that 1 GB RAM simply did not exist. Later it did exist. But was too expensive. Indeed it continued not to exist at the needed cost until 35 years later and then suddenly there was enormous progress in AI. It was not from nowhere.
So while it became in properly some ways easier for a new app to become one of the most widely used apps, it meant almost all time is spent in only a few dozen apps total, whatever they are in that epoch.
Comment. Usually whenever somebody has a great idea, people, if they agree it's good, also suggest it may not be all that new. Compare this with what happened with AI. Due to how late the merely possible because the actually feasible. AI made enormous progress as if suddenly. The ideas were most often so old that most people thought they were wholly new.
Summary. After they noticed (a), (b), (c), many people began to imagine that low hanging fruits in software were running out. They began to entertain the dark thought that, maybe, possibly, all the low-hanging fruits in ``tech’’ were simply taken and eaten in the 90s, 00s, 10s, and that now there aren't any more such low-hanging fruits.
Comment. There are always many business plans and other technologies that are not used only for lack of sufficient capital. We are not talking about that.
However, the truth might be rather the following. There are countless many low-hanging fruits remaining in software. Rather it was the social low-hanging fruits that may have been picked, taken away, and eaten.
Silicon Valley and more broadly California defined the success of many technology startups in the 70s, 80s, 90s, 00s, 10s, simply because enough investors there largely, it seems, acted much like a single, organized social network.
Therefore it was far easier to be given resources needed to make a new startup succeed than in other places. In other places, one outside investor would join the team, and that would be all for the round. To make proper contact with another outside investor required travel, navigating a completely different social network, making new friends, all costly, difficult, and worse, the main cost was time. California meanwhile was largely one social network to some extent. Lower costs, bigger rewards, sooner, with less risk.
Searching for more capital and growing a great product sooner and faster is often much more difficult and time consuming than actually making a great product or a demo. Perhaps it is even more difficult than getting a large number of users.
Everyone now sees that some low-hanging fruit aspect is increasingly missing today. People merely have a hard time saying which one. It seems much harder to make a startup and succeed as quickly as previously.
3
But here is the main point: communication amplifiers were not actually the only trend in the future of software. There was a larger trend. People didn’t quite notice it, because it was apparently obscured but its most prominent component instance.
Communication amplification was part of that larger trend. But it was not the whole trend.
It is the larger trend that appears to have contributed to the greater, easier success of the electronics industry compared to heavy industry, to the greater, easier success of the software industry compared to electronics industry, to the greater, easier success of the data-is-the-product industry compared to software industry, and that is contributing, presently, it seems, to the greater, easier success of the artificial intelligence industry compared to data-is-the-product industry.
Take a guess what is that trend?
Hint: it has to do with lower and lower risk.
The trend is a movement away from producing something that has lower, narrower tolerances to something that has greater, wider tolerances.
Ultimately the trend appears to be a movement away from specifications entirely.
For example, what search result is a ``good’’ or ``good enough’’ search result? And more users accept ``good enough’’ search results than accept ``good enough’’ food in a restaurant. Yet people also precisely define ``good’’ and ``bad’’ food in a restaurant. Or a ``good’’ batch of material.
The lower the tolerances and the harsher the specification that must be satisfied for the product to work or to be acceptable and be sold, the more risk there is in the business plan and the fewer resources available for rapid growth.
For example, heavy industry produces one batch and then another. There is always the possibility that each batch might not have all characteristics fall inside the tolerances and be rejected.
Often the tolerances are very tight for natural reasons, not merely social convention and consumer preferences, required for the product to work satisfactorily, and then even slight deviation would significantly reduce the value of the batch.
Each batch where even a single parameter, out of a dozen or more, deviates even a little from a definite range most often must be thrown out and the cost swallowed by the manufacturer. Yet most such parameters vary independently. Each parameter is also never completely in the control of the manufacturer. Too often adjusting one parameter disturbs some other parameters. Several batches must be sold with profit to balance the loss of a single batch that cannot be sold and whose cost must be swallowed by the manufacturer.
This gives heavy industry, and really most industries, most human production, one of the least desirable possible overall patterns. Heavy industry and most human production has the pattern of (A) many small gains and a few large losses. Whereas a much better overall pattern is (B) many small losses and a few large gains.
Surviving and growing with pattern (A) is much more difficult than surviving and growing with pattern (B).
Comment. Obviously no losses at all would be best. But there are always parts of the world that we do not control and losses can always happen. The only real question is what is their pattern. When do they happen.
Digital electronics, software, games, data-as-the-product, artificial intelligence, seem to be part of a larger trend away from working in industries with business pattern more like (A) and rather working in industries that have a business pattern more like (B).
Humankind has mostly worked with pattern (A). Almost all production happens to be like that.
Most production makes a physical product. There are many requirements that are independent and yet the value of each batch sensitively depends on the precise properties of each batch. Almost as soon as new industries appeared, due to advances in technology, industries making products that could have more leeway, people rapidly moved to work in those industries. And overall growth in those few industries was much faster and greater than in all other industries.
If they think about it, or even if they merely imitate success, and if the primary concern is greater probability of survival followed by more-rapid-than-is-common growth, people seek to do work that has less pattern (A) and more pattern (B). This creates a direction. Every success then reinforces that direction.
Products that amplify communication are outstanding example of ``low spec or no spec’’ businesses with enormous tolerances.
More users is often sufficient to increase the value of a platform. Its value does not vary sensitively with the precise features or how those features are precisely implemented. The value is often from the data generated by the activity of users in such platforms.
That trend appears to be the overall best means of predicting not merely the future of software but of all human industry. Human activity looks not merely toward communication amplification, but toward industries where product specifications and requirements play a less and less significant role. It appears to have been the trend for over a hundred years.
Opportunities to do something whose pattern of outcomes is less like (A) and more like (B) are exceptional in human history. They tend to create exceptional growth, new markets, new corporate giants, and establish a direction, a gradient.
Hit the like button. Subscribe now.
(First draft archived also at: https://telegra.ph/Several-trends-in-game-and-software-development-01-20)
(Second draft archived also at: https://telegra.ph/To-predict-the-future-of-software-Part-1-03-25)
Disclaimer. This work is licensed under a <a rel="license" href="http://creativecommons.org/licenses/by-sa/4.0/">Creative Commons Attribution-ShareAlike 4.0 International License</a> ©Len Alexander. This text is a popular, speculative discussion about a subject matter that interests the author. It is solely the personal opinion of the author. Consider it as mere conjecture, provided for entertainment. Regarding it no warranties of any kind exist. No promise to do anything or that anything in particular is or was done is present.