Vannevar Bush’s “As We May Think” is a 1945 essay on the future of information management. In it, Bush makes eerily prescient predictions about how technology can empower human cognition. It’s easy to see how his writings about the Memex (a theoretical device that can store virtually unlimited amounts of information and surface related documents – basically a personal computer) have influenced the development of technology over the past 70 years. While we have made great strides in enhancing access to information, the problem of information overload is still one of the largest facing us today. Now, we face a different aspect of this problem than we did pre-internet and pre-Google.

The internet and search engines like Google and Bing have made information extremely accessible to the average user. However, search engines haven’t completely solved the problem of information overload. Search engines help us find relevant documents, but there still may be too many relevant documents for a person to process comprehensively. We need to come up with new ways of aiding humans in processing and finding salient information in the vast amount of information on the internet and in our computers.

Bush talks about enhancing the human intellect in his discussion of information indexing vs. association. Computers operate by ordering information into indexes. Given an ID, computers can easily fetch the corresponding piece of information. However, humans are more likely to operate by association. We don’t easily remember IDs but instead recall concepts, themes, and keywords. It’s for this reason that billions of people go to Google instead of typing URLs directly into their browser. How can we extend upon the concept of the Memex and allow technologies to better enhance human thinking? In this post I discuss a couple avenues of interacting with information that I think are worth investigating.

How will we think?

Bush notes that when thinking of the Memex and other ideas, “technical difficulties of all sorts have been ignored, certainly.” As an engineer, it’s hard not to think of technical limitations, but I’ll try my best to think of ideas that aren’t restricted by what’s technically possible. Here are some of the technologies that I’d like to see in the near (and not so near) future:

A truly personalized search engine.

This service would have the ability to ingest all of your emails, text messages, bookmarks, and documents on your computer in order to learn your personal corpus. Then, as you’re browsing new webpages or documents, it would suggest relevant documents that have already been indexed. There would probably need to be an emphasis on encryption and privacy because of the personal documents being indexed. Furthermore, ranking may be more difficult because you would be the only user contributing data to the search engine. This could open up better ways of knowledge sharing/transfer in institutions (e.g. someone who’s retiring could give their successor their knowledge base to ease the transition).

An actually smart personal assistant.

I spend a lot of time writing notes and save a lot of bookmarks but rarely take the time to go back and consult them. Why limit information retrieval / search to a process initiated by the user? It would be useful to have a smart assistant that has access to your personal corpus of documents that could surface salient items while you’re writing a paper or reading a blog post. That way, information would be offered to you without you having to break your flow and go and search for something. Bush writes about the Memex having the ability to “stop only on the salient items” and how “a new profession of trail blazers, those who find delight in the task of establishing useful trails through the enormous mass of the common record”. Why not automate the trail blazing?

Brain computer interface enhanced information.

It’s amazing that Bush was thinking about brain computer interfaces all the way back in 1945. As for the relevance to information retrieval, perhaps BCIs can help provide better data on the relevance / interest of a document to a particular user. If click-through rate is currently used as a measure of how relevant a user thinks a document is, imagine being able to see a user’s brain activity as they’re reading a document. Maybe the search engine could check to see if the correct brain regions are activated and use that to improve ranking algorithms. Also, imagine being able to detect what a user is thinking about and then automatically offer related documents / results as someone is thinking about a topic, similar to the idea above.

Furthermore, I wonder about the ability of BCIs to modify, and not just read, brain activity and how that could enhance the depth of information. Could feelings be extracted from one person, stored on the internet, and then shared to another person by encoding, transmitting, and releasing specific neurotransmitters? What about memories? Can brain states be shared? Could one person’s specific feelings of curiosity or interest in a topic be transferred to another? This would add a whole new, less quantifiable, dimension to shareable human knowledge. This also seems like it’d be far beyond 10 years of technology and also could open up a whole can of philosphical worms.


Overall, this was a fascinating read and I wouldn’t have been surprised to find out that it was written in 2005, but am shocked that it was written in 1945! It’s sort of eye opening how fundamental the problem of information overload is and it’s amazing that people were thinking about it in 1945, before digital computers and the internet.