Inside Steve’s Brain and “Vectors going in time.”
I read Inside Steve’s Brain this weekend. It was a quick read, although the book isn’t one that I would recommend to most of my friends (except the extreme Mac-heads out there). However, there was one point that I wanted to explore further — Kahney discusses how Steve Jobs pays keen attention to the convergence of various technology trends (which Jobs calls “vectors going in time”) to do things that were not possible previously.
The iPod was made possible by cheap, small storage (1.8 inch drives from Toshiba), the proliferation of digital music (ubiquitous PC CD burners helped to make that possible), broadband penetration in the home, and advances in the manufacture of plastics and later metals. YouTube was made possible by ubiquitous availability of Flash in the browser, cheap storage and bandwidth, and broadband penetration in the home and workplace.
So what are some interesting “vectors going in time”?
There are a number of interesting trends that have the potential for major disruptions. Here are a few that come to mind (please contribute your thoughts in the comments of this post):
+ Cheap and portable storage.
Flash memory is going to change the game — by both reducing the space required for storage and because flash memory is so durable. What devices can be re-invented by leveraging Flash memory? On the other end of the spectrum, storage capacity (of the spinning disc variety) is growing at such a rapid pace that it’s just about free. What can you do with infinite storage capacity?
+ General purpose Graphics Processing Units (GPGPU).
GPUs are brilliant at offloading (from the CPU) highly intensive graphics tasks. Graphics is an inherently parallel task (every pixel is separate), so using many GPUs to handle computation is a superior solution to jamming everything through the CPU. With data volumes exploding, are we finally ready for parallel computing? Are there more tasks ideally suited for this technology? Will something emerge that offers a layer of abstraction that allows the massive investment in existing programming techniques to be sent to a cloud of GPGPUs?
+ Distributed file systems + MapReduce.
Projects like Google’s GFS/MapReduce and the open source Hadoop are frameworks for running applications on highly distributed commodity hardware. Google’s entire search index was rebuilt on GFS/MapReduce as soon as the software was ready. These frameworks are critical in order to process petabytes of data. They are also relatively new and improving at a rapid clip. How can we take advantage of Hadoop to do cool stuff with huge data volumes?
With infinite storage capacity and compute cycles, what can we do with all of this data? Kevin Kelly wrote a great piece called The Google Way of Science. Machine learning is going to evolve beyond it’s limited use today in bio-informatics, web search, and anti-virus products. What really hard problems can we attack today with machine learning which were previously impossible?
+ Ubiquitous wireless access + AJAX.
Just as fixed-line broadband created many new opportunities, so too will the ubiquitous availability of wireless — 3G, GPRS, CDMA/GSM, WiMAX, Wifi, or Bluetooth. This plus the move to AJAX (asynchronous Javascript and XML) will finally allow server-side computing to own the planet. What applications benefit most from mobility?
+ GPS everywhere.
GPS will soon be a default in all phones. The cost of GPS will eventually be low enough to embed it in all devices. With wireless everywhere, devices will both be able to tell us where they (and we) are AND communicate that information to server-side services real-time. What services would be dramatically better by having locational information?
+ New human-computer interfaces.
A friend of ours picked up a new iPhone — her four year old daughter picked up the phone when she wasn’t looking and started using it. Not the way a four year old uses a Blackberry, but really using it like an expert. The touch screen interface is such a natural way for humans to interact with machines.
The Wii has taken the lead in the console wars with inferior graphics processing capabilities — the reason because Nintendo is innovating on a different vector — human-console interactions. The Wii controller’s gyroscopic interactions with the console and Wii Fit’s accelerometer has a nailed virtual reality. How can we use these emerging interfaces to build better web experiences? The iPhone also has an accelerometer. Hmm….
6 Comments
July 1, 2008 at 9:56 pm
Mike - In my opinion some of the most interesting “vectors going in time” aren’t related to technology at all, but rather relate to human behavior. The one that is most interesting to me is how consumers have moved their “Cognitive Surplus” time (term from Clay Shirky) away from TV and onto the web in a participatory fashion. It is really this shift that has enabled wikipedia, youtube, etc. There is a huge opportunity for companies to take the concept of the cognitive surplus and combine it with the technical “vectors in time” you outlined to build interesting products & businesses.
July 2, 2008 at 6:39 am
Great post, Mike.
I agree with Mark Williamson’s comment. There are interesting behavioral vectors as well as technology vectors. I would add self service and green lifestyle as behavioral vectors.
On the technology front, there are a couple of other items I’d suggest for your list:
-service oriented architectures with loose coupling, non-ACID semantics, and non-deterministic orchestration.
-search as a fundamental human and application interface.
-XML as a representation and transport protocol
-large scale parallel/scale out computing (the distributed file systems you note above are a subset of this broader technology trend)
-increasingly powerful, smart and ubiquitous mobile communications and computing devices (again overlaps with your wireless point)
-adaptive/autonomic computing architectures and systems
Finally, I would suggest that there are a couple business model related vectors as well:
-open source
-software as a service
July 2, 2008 at 9:12 am
There’s another technology vector that I’d like to add to the list in my previous comment - virtualization. Computing and storage are increasingly presented as virtual resources.
July 3, 2008 at 8:53 pm
Interesting vectors I see as an infrastructure strategist:
- peak off-load computing: EC2/S3ish ideas
- derivative enterprise computing costs: security, redundancy, DR, audit, management software, appropriate personnel > all making the case for consolidating in best practice pools and sharing costs (aka data centers)
- mobile computing data loads: swarming real-time data, not only GPS, but include accelerometers, thermometers, etc… aggregating, analyzing and regurgitating that data will create very interesting data services — what’s the infrastructure required for that? (a real-time Hadoop engine?) (just add execution…)
July 7, 2008 at 2:56 am
I’m quite psyched about the upcoming availability of AJAX-everywhere … especially with jQuery bridging the gaps.
Also keep one of your eyes on neurofeedback … it’ll synch your brain up to your handheld device.
July 7, 2008 at 4:44 am
@ scoot fitchet, definitely excited about hooking up neurofeedbac. Touch typing was a huge leap for me, because it’s so much faster than writing, but it still doesn’t approach the speed of thought. And I think neurofeedback devices are closer than we think.
The same goes for machine learning–I’ve seen some very impressive feats in AI recently. I can’t wait to see what the next decade brings.
Leave a Reply