The internet today is structured so that your personal digital data is stored, owned, and used to enrich a select few strangers who control this. We default to that structure as the status quo. But if you allow yourself to think about a fully decentralized internet, a compelling vision of equality and goodness unfolds.

A person’s digital footprint already exists at birth. Then each day we generate more digital data that steadily grows as we age until ultimately we leave it all behind one day as our digital shadow.

Of course, there are varying degrees of how connected people choose to be. But it’s increasingly enticing to participate in an internet-connected life with our phones as our primary gateway. The collection of digital data we generate in our lives greatly impacts our relationships, education, career & finances, entertainment, and opportunities.

I believe the data is sacred to the individual as a near literal extension of one’s person. The data is perhaps even more valuable to our lives than our actual shadow. Maybe a better analogy would be an organ of our body. Organs enable us to do things when healthy and cause pain and suffering when not. Organs can be harmed by bad guys and must be protected.

Today’s internet treats our data with the respect of a shadow; we need to start treating it instead with “organ-level” seriousness. The premise is that you should enjoy better control and assurances of protection and stability for your data and be able to make money from your data if you choose, and otherwise have full decision-making control, if you allow it to be used in the training of AI apps that others build. If you have to approve every AI model that accesses your data, the economy of tech might rightsize itself with the rewards of the economy, flattening back out more equally among us.


To date, we have accepted an internet that stores our personal data in centralized server farms controlled by a handful of tech CEOs. Every day we give them more of our personal data. We click license agreements without reading them, lost in their legalese and hopeless to their whims. Our individual data is just one node in their computing enclave.

While I admire the entrepreneurial journeys of the tech world, I increasingly am concerned that we’ve built an ethically wrong structure. It clearly places inordinate power over billions of lives in the hands of a few and reeks of oligarchy. While this power structure is easily recognizable as inappropriate, the fact that we’ve submitted our data to these centralized servers, accessible only by the oligarchy, is further problematic in that the oligarchs have a wildly unfair competitive advantage building future AI products since the ability to build good AI projects is dependent upon access to vast quantities of good training data. Since the oligarchy has our personal data, they can continue to build products that no one else can, not because they are smarter than the rest of us, but because they are just holding our personal data.

For these reasons, it is worth brainstorming solutions that bring data physically closer to the individual, maximizing freedom, control, fairness, and value for every individual.


Decentralization is the process by which centers of power can be broken up and distributed back to individuals. For example, due to blockchain trust, cryptocurrencies are showing how decentralization can break up centralized power in the financial sector. Can similar decentralization principles be extended more completely to all aspects of internet life?

From Wikipedia, “Digital footprints are not a digital identity or passport, but the content and metadata collected impacts internet privacy, trust, security, digital reputation, and recommendation.” Let’s envision how a hypothetically decentralized internet with a decentralized personal datastore would impact each of those areas. Let’s assume the datastore is governed by a healthy open source foundation with a noble mission that includes making decisions that benefit individuals, one equal to another, over any other consideration. Let’s assume we have the same benefits of crowdsourced attention to privacy, security, and features that have been exemplary for instance in the development of Linux. Openness eliminates secrets and ensures accountability.

  • Internet Privacy:  having complete control over your data, you can limit what anyone else does with your personal content; laws will naturally adjust to the new paradigm, likely protecting personal datastores with the similar protections given to other personal and property rights
  • Trust:  since everyone has equal access to build the same personal datastore and since everything operates on principles of fairness and openness, trust can be given space to grow
  • Security:  other open source efforts have shown how to be successful with security, and a decentralized internet would benefit from the difficulty hackers would face in tracking down datastores scattered around the world
  • Digital Reputation:  through the elimination of information gatekeepers that arbitrarily modify digital reputation, perhaps a decentralized internet would be able to innovate better ways of maintaining peaceful and healthy relationships online
  • Recommendation:  once people have control of their own data, they will innovate ways in which individuals can accept payment in return for allowing their data to build AI software; potentially allowing them to reap the benefits of a really good AI-enhanced experience online if they so choose

With a strong vision like that, I wonder if this is actually something that we could pull off. I think so. It’s definitely an interesting thought exercise. And since you’ve made it this far in reading my post, perhaps you too can come up with interesting ways in which we can improve this vision or, if there are mistakes, call them out.

I am thinking through bold plans to help tip the internet appetite further towards a decentralized future. I’ll write more on the topic soon. What do you think? Are there realistic conditions that could compel this sort of future?


(Side note: Congratulations today to Brian Armstrong and the Coinbase team on their IPO — a remarkable decentralization moment today!)

{ Comments on this entry are closed }

Take Me Away From the Norm

October 6, 2020

in AI

I posted on the ArrayFire Blog earlier today about advancements at NVIDIA and ArrayFire, moving forward the blossoming field of AI. Read that post first before continuing.


I found a song that captures the AI movement, though the song actually symbolizes much more. Additional meanings I leave to the reader.

To add to the mystique, 3:11 was the timestamp of the video from the ArrayFire post. Listen to this song.

Lyrics are here.

To appreciate it from a computer science perspective, think of “brainstorms” as AI. Recall from the intro GTC video that AI is “a composer of music.”

Let’s have fun. You might consider that the builders of AI are “blind to fear” and that we need to follow the advice, “Don’t give up your independence unless it feels so right. Nothing good comes easily; sometimes you’ve got to fight.”

Then it becomes more practical with “launching a thousand [threads] in [its] heart,” the GPU, or other data-parallel devices.

Finally, it wraps up with, “Whoa, [AI], take me away from the norm. Whoa, I got to tell you somethin’.”

At ArrayFire, we’re working with AI, and we’ll tell you more things soon as we reboot this blog and the ArrayFire company blog!

{ Comments on this entry are closed }

Signing NDAs

July 15, 2015

Eight years ago, as a new inexperienced entrepreneur, I was excited when a BigCo sent us our first NDA to sign. Another company actually thought we were important enough to get their legal team involved. Haha! I was also nervous. What if they were trying to pull one over on us?! I’d better get our […]

Read the full article →

Firing Clients

July 14, 2015

One of the toughest things to do is to fire a client. Sometimes a particular client may pay their bills, but they may cause all sorts of havoc for your startup. We have been unsuccessful at purposefully firing clients. I am not sure if our decision-making has been right or wrong, but these are some […]

Read the full article →

The Overhead of Doing Business with the Federal Government

July 10, 2015

The first 6 months of this year, we have had to introduce multiple type of overhead into our operations to do business with the federal government. Essentially, in order to receive a larger contracts from the federal government, a business must have certain accounting standards. This is called being DCAA compliant. The biggest difficulties in DCAA compliance are: […]

Read the full article →

Computational Speed Needs of the Federal Government

July 7, 2015

ArrayFire is about making code faster. Many of the problems of the world that need to be accelerated are managed by the federal government. We currently are working on two major projects. We work with Defense Advanced Research Projects Agency (DARPA) which is part of the Department of Defense on the XDATA and MEMEX projects. […]

Read the full article →

It’s a Numbers Game

July 6, 2015

Recently I have thought a lot about the phrase, “It’s a numbers game.” A lot of success in life is all about repetition. Many people perceive others’ success as being lucky where in reality the success may have been the probabilistic result of taking more shots on the goal. Here are some scattered thoughts about this: […]

Read the full article →

Empowering Leadership

July 4, 2015

Yesterday I wrote about how a change of leadership style affected our team. I could not articulate exactly what had happened until I saw this video. The two leadership styles depicted in the video are the two conflicting styles that my employees saw. My personal style has always been light and places the burden of decision-making […]

Read the full article →

Startup Integrator Fail

July 3, 2015

Last year, I played a hunch and hired a startup integrator. I thought our business needed one, and it is an idea from a well-known book called “Traction”. The goal was for this individual to run the day-to-day with”in” ArrayFire while I worked “on” ArrayFire and “on” our two new startups. The mandate for the integrator was […]

Read the full article →

Killing Off Brinktop

July 2, 2015

Last summer I attempted to start two new business:  Dyced and Brinktop. I have already written about the rise and fall of Dyced. Dyced lasted only 2 and a half months. Brinktop on the other hand lasted nearly 10 months. I introduced Brinktop before. Brinktop started to make highly customizable laptops. But laptops are priced too low and […]

Read the full article →