Slashdot Mirror
Marc Andreessen Describes Vision of 'Ambient Computing' (telegraph.co.uk)
An anonymous reader writes: Marc Andreessen, co-founder of Netscape, is one of the biggest investors in technology. In an interview with The Telegraph, he spoke about how he envisions the future of computing. It's essentially an extension on the idea of the "Internet of Things." He thinks mobile phones will begin to be replaced in just 10 years. "The idea that we have a single piece of glowing display is too limiting. By then, every table, every wall, every surface will have a screen or can project." Within 20 years, he expects most new physical objects to have some sort of chip implanted within them. "The end state is fairly obvious — every light, every doorknob will be connected to the internet." The term for this is "ambient computing." There will obviously be a transition period — perhaps the so-called internet of things is just an early phase of that transition. But with powerful chips and sensors becoming incredibly cheap, Andreessen's scenario seems possible. I guess it's time to get cracking on those security and privacy concerns.
There is already a domain where ubiquitous integration of high complexity capabilities into virtually all materials with room for them is a reality.
We call it "Biology". And, in my professional capacity as a fungus, let me remind you that you'd be fucking insane to want your computers to go down that path.
It's impressive that machines made of meat work at all; but that doesn't change the fact that they are tottering heaps of uncontrolled complexity, riddled with pathogens and parasites, kept alive only because they are (sometimes) more fault-tolerant than they are faulty; and because the various microorganism militias are too busy fighting assorted cryptic metabolic battles and it is possible to enter alliances of convenience with some of them, if you get lucky.
People have done a terrible enough job keeping a bunch of loosely-linked deterministic finite state machines from descending into a putrid jungle of malware that inspires comparison to unpleasant biological outcomes. You want to add more; and link them more tightly? Have fun with that.
That line edited for the summary (and people say editors don't do anything) actually goes like this:
"The idea that we have a single piece of glowing display is too limiting.
By then, every table, every wall, every surface will have a screen or can project," he told the Telegraph.
"Hypothetically you walk upto a wall, sit at a table and [talk to] an earpiece or eyeglasses to make a call. The term is ambient or ubiquitous computing."
If tables, walls etc. have screens or projectors - why special earpieces or glasses?
If special glasses are ubiquitous - who needs screens in tables and walls? Wouldn't empty and clean flat surfaces be far more useful then?
And what's the use of a chip in a chair? To tell you that it is occupied or not? Wouldn't a single camera in the room do that and more?
Article further gives examples such as:
Pharmaceutical companies transporting drugs or vaccines need to constantly monitor temperature;
logistics or delivery companies track their fleet of vehicles over long distances;
and perishable food companies need to monitor internal temperature and humidity of trucks to check if their goods are spoiling.
None of that is "Internet of Things" or "ambient or ubiquitous computing" nor would those examples benefit from chips and sensors on every single item that is being transported in those trucks.
Why then all those additional sensors?
Well... cause it would be expensive to make dedicated case by case monitoring systems to replace somewhat manual (but cheap) solutions already employed. As in, there's an employee doing that right now.
So the solution is to cover EVERYTHING with sensors instead.
Thus eliminating the cost of installing sensors and networks by shifting it to every producer of everything in the world.
Who would then shift that cost (and all the unnecessary "features" they'd have to invent for their products) - to the customers.
While "the cloud" will pick up the rest.
Thus, "reducing the costs" of creating "ubiquitous computing" to software only - i.e. nothing, as developers are already being paid anyway, or they'll just do it for fun and experience.
That's the logic.
"The problem is that manual measurements are very common in hospitals, pharmaceutical delivery chains, and even the distribution of dairy and meat produce.
Someone actually goes to the warehouse to fill out a report with pen and paper every 3 hours," says Samsara's CEO Sanjit Biswas, whose previous network technology startup Meraki sold to Cisco for over $2 billion.
His big idea: installing cheap sensors, and uploading and analysing data to the cloud makes Samara 1/10th of the cost of existing industrial sensors (complex systems made by huge incumbents like Intel), and deployable in under 10 minutes.
"If you want a tailored system, someone like IBM will build you a custom solution but it usually costs $5m so it doesn't make sense unless you're a large company," he explains.
Andreessen is a fierce believer in the impact of this wave of software-driven sensor startups.
His core thesis is that over the next 20 years every physical item will have a chip implanted in it.
"The end state is fairly obvious - every light, every doorknob will be connected to the internet.
Just like with the web itself, there will be thousands of of use cases - energy efficiency, food safety, major problems that aren't as obvious as smartwatches and wearables," he says.
Except that is not "software-driven sensors" but "sensor-driven software".
Which relies on someone first providing ubiquitous sensors in every doorknob - which could then be used for "major problems that aren't as obvious".
I.e. It's a solution we don't really have a problem for quite yet. But it would be great if someone else paid for it.
Mit der Dummheit kämpfen Götter selbst vergebens