Spreadsheet—a personal history

When I was a kid, I created a game of cricket in a spreadsheet program called Lotus 1-2-3. Each die roll would represent a ball being bowled in a cricket game, and the number told me what happened —a batter got out, or scored a run. I kept rolling the die and entering the numbers onto the spreadsheet and in course simulated a game of cricket. My cricket universe within the spreadsheet tracked statistics—batting averages and strike rates for all the players. I spent an equal amount of time playing the game and improving the spreadsheet.

The first evidence of humans using something akin to a spreadsheet comes from ancient Sumeria. Archeologists have found stone tablets etched with tables tracking some form of accounts. Accountants before the advent of the computer, used to maintain ledgers that were spreadsheets on paper. These had one major disadvantage over the electronic version—if you had to rewrite, recalculate or etch over the entire table if a single value changed. If you were maintaining the accounts of a coffee shop and the price of coffee went up that meant days of tedious boring work to create a new copy of the sheet.

Electronic spreadsheets solve this by using a concept of “references”. A cell can contain a formula that references a value stored in another cell. So if you update the value of the cell containing the price of coffee beans, the whole spreadsheet updates instantaneously. This was the key insight in the predecessor of the modern spreadsheet called LANPars. It was a programming language that would output a sheet.

VisiCalc was the first incarnation of the modern spreadsheet in a design we recognize. VisiCalc was a clunky program with green and black flickering cells but it had the numbered rows and alphabetical columns that we now find in all modern spreadsheets. VisiCalc was a raging success. People bought the two thousand dollar Apple II to run the hundred dollar VisiCalc. Small stores started to do their accounts in VisiCalc. Dan Bricklin the inventor of VisiCalc was actually worried that the accountants would soon be out of a job.

As a professional software engineer my first duel with spreadsheet was when I was working at Bright—a solar energy startup that was selling solar panels in Mexico city. The company had been around for six months when I joined. Pablo the head of operations, built the company by using spreadsheets as duct tape. To generate a quote sales people would enter some numbers into a sheet and it would spit out the price, the engineer on site would record the size of the roof on another sheet and that would trigger an alert to our back office where folks started procuring panels of the correct size. At that point, I did not have a lot of respect for the spreadsheet, I thought it was clunky, error prone and a tool suited for non programmers. I had just heard about the Blue whale fiasco at JP Morgan London where an incorrect formula in a sheet lead to a wrong risk calculation of the deal that lead to a loss of a billion dollars for the bank. Pablo complained to me that the spreadsheets had become too slow. I started to build a website to provide the same features as the sheets. Each sheet took me two weeks to build out. I had to reinvent things like the capability to upload files and display PDFs that Google sheets provided out of the box. I had eleven of these sheets to copy over. As I was working on the problem, sales wanted to change the formula for discounts. Everyone knew how to modify the spreadsheets, but only I could change the software. That’s when I realized how accessible spreadsheets had become, almost everyone knew how to use one and it helped them get their job done faster. Spreadsheets were increasing productivity for everybody.

Back to the Apple II, the VisiCalc was so popular that in typical silicon valley fashion competitors started to clone it. The most successful clone was Lotus 1-2-3. It was named 1-2-3 as it was “as simple as one two three”. It was a highly polished version of VisiCalc, less clunky and could take advantage of the larger memory available on newer computers and it would run on the upstart operating system—Microsoft DOS. To top all of this, Lotus 1-2-3 was the first software to be advertised on TV. VisiCalc stopped selling overnight. But the buccaneer spirit of the valley was not to be satiated. Borland, another software company, released a competitor to Lotus 1-2-3 cheekily called Quattro or four. To lure users from Lotus, Borland built compatibility mode that would let Quattro operate Lotus files, it even had the same menu options in the same order as Lotus 1-2-3. This infuriated lotus and they went to the court. After six years the court decided in favour of Borland, by providing an analogy to the VCR—“The buttons on a VCR are used to control the playback of a video tape, similarly menu commands are used to control the 1-2-3 file. Since the buttons are essential to the operation of the VCR their layout cannot be copyrighted.”

My next major encounter with the spreadsheet professionally was when I was working for a shipping logistics company. I was the lead of a team building software to predict the price of shipping a container on a route—say from HongKong to Los Angeles. The benchmark to beat was a spreadsheet created by one of the Vice President’s of the company. The legendary spreadsheet and the acumen of its creator was one of the companies competitive advantage. The software we built was compatible with spreadsheets from day one. Business associates could update the prices by uploading a spreadsheet into software. They had a manual mode to always override the software generated values. This time I deferred to the powers of the spreadsheet. My lesson here was it is foolish to try to replace the spreadsheet everywhere, it does a lot of small things for a lot of people in a lot of different ways. People have all these neat automation tricks in their spreadsheets that they do not tell you when you do user interviews to build a better replacement for the sheets. It’s second nature to them, they are familiar with the interface, they know the keyboard shortcuts and it is fast.

The Borland vs Lotus fight did not matter much because Microsoft soon released Microsoft Excel. By the time version 5 of Excel was released in 1993, bundled in the office suite, it was the clear market leader. For the next fifteen or so years, Excel and spreadsheets meant the same thing. People started using Excel in ways its designers did not envision —schools were using it to arrange seating layouts and doctors were using it to calculate the correct dosage of anesthesia before a surgery. The accountants did not lose their jobs as more businesses wanted to simulate many “What If” cases—what if we hire 100 more people, what if we give out a smaller bonus, what if we acquire that company. It was the finance industry that Excel transformed the most. Deals were made by exchanging models built in Excel. Today there is an Excel world championship where analysts duke out to find who is the best at using Excel.

The next big jump in spreadsheet technology came around when Google launched the online version of it by cobbling together technology acquired from a few startups. By 2010, Google had made the spreadsheet collaborative—multiple folks could edit a single spreadsheet in real time—and on top of it Google made it free. Pablo’s spreadsheet process only worked because multiple people could update the information about a single customer in real time.

Joel Spolsky who was a famous program manager for Microsoft Excel, claims that in his user studies he saw that the most common use case for Excel was to record lists. During the me too movement, a clerk at one hollywood company created a spreadsheet list of media men who had abused women. This list was updated by women anonymously. This led to investigations and sometimes firing of senior media mein in firms like NPR, Buzzfeed and New Yorker.

It is fashionable to say that “software is eating the world” in Silicon valley. Spreadsheet is leading the charge on this front by finding uses in all industries. Software now is chewing on what Excel has already masticated. Nate Silver used Excel to create the models and the charts to usher in the new field of data journalism. Spreadsheets are the de facto tool to collect data in academia. A recent paper analyzed three thousand genetic papers and found that one in five papers had an error due to a common error—the gene “Septin 2” which is shortened to “SEPT 2” automatically becomes “2-Sept” in Excel and the whole calculation goes for a toss. There are full RPG games made inside of Excel. Gym trainers share workout routines by sharing a spreadsheet. Casinos use it to optimize slot machine layouts.

I use spreadsheets to plan my finances, to coordinate bachelor parties and to create fancy charts to make a point at work. The creators of the spreadsheet did not build a program to solve a single problem, they built a general purpose tool that improved productivity. Spreadsheets made money for Micorosoft and other companies but provided a lot more value to society. Good software is never a zero sum game.

Spreadsheet is a powertool. Its design has been copied over many times by different companies. In a world where the trend seems to be about building good looking single purpose software that scales well perhaps the option of building general purpose tools that look and feel the same across industries is an opportunity worth investigating.