Steve Ballmer is the former CEO of Microsoft and the current owner of the Los Angeles Clippers. He is also the creator of USAFacts, a website  dedicated to organizing and presenting data that allows U.S. citizens to understand the population, finances and government policy of the United States.

I interviewed him two weeks ago about his project. The interview has been split up into two parts. The first part (below) focuses on USAFacts. The second part (to be published Friday) also discusses broader questions about the role of facts and data in the U.S. today. Both parts have been edited lightly.

HF – Previous interviews have suggested that your site saw a big burst in initial numbers, and that things have tapered off a little since. What do the figures look like at the moment?

SB – We’ve had a total of about 750,000 unique visitors, which is great. Every state in the union and almost every country in the world. That part’s great. I would say that we’ve tapered down to about 3,000 visitors a day, which is not many, in my opinion. But every time we get out and talk about the thing, we see a surge.

I think there are probably three or four different things going on. First, we have to work harder to make our site findable. Second, we have to make our site more consumable. Third, we need dialogue and discussion in general on the value of data and facts in decision-making. Whether people are coming to our site or going somewhere else, the key for us is to get people to really look at the numbers and look at them in context.

A lot of what people like to do is to pick numbers at random, so we think it helps to have something that is contextual, complete and comprehensible. We think that we’ve done a really good job in making ourselves comprehensible, but it could really be better. I think that people like the facts, but not everyone loves numbers. Bringing those things alive in graphical form will help.

The problem with words is that they have a harder time capturing context and being accurate. When you say that something is very small or very large, I don’t know what that means. If I say that this number is 10, and this is 11, that tells me what you mean by very large and very small. Adjectives and words are wonderful things, but at least for the part of government that we’re focusing in on, the part that can be described numerically, we think it’s really important to use the numbers.

HF – So you are planning to work with data visualizations?

SB – Yes. We have some cool ideas for that, using — no surprise — Microsoft’s Power BI technology.

HF – So do you have any idea of the profile of the 3,000 users who come every day?

SB – They tend to be younger. They come disproportionately, which may not be surprising, from the state of California. They come from New York and D.C., and also the state of Washington. There is a strong tech, educated community reflecting not only Microsoft but Amazon and the start-up community. Those who use the site tend to be younger, by and large, and a little more liberal than average, but not by much. Other than that, they are a pretty representative set of people.

HF – Have you plans to create an API [an interface that allows other programs to easily access data that they can then re-use and repackage]?

SB – We will create an API. We don’t have one now for two reasons. First, some of the data is still in files that are really hard to access. We’re putting everything in a SQL [structured query language] database — we’ve been working on this project with a team at the University of Pennsylvania who are experts both in data and development. We need the database complete, and then we’ll test the API to make sure that it’s scalable enough to handle the volume that we want. That is absolutely on our road map.

HF – Have you seen people mixing or using the data in unusual ways?

SB – The most interesting was Outside Magazine. They said “Hey, we care a lot about the great outdoors. Let’s take a look through the data to comment about how government’s involvement in the great outdoors has evolved over time.” I thought that was cool, and it shows some of the kinds of things that people can do to create their own views of the data.

HF – You talked in a previous interview about how numbers can end arguments. Yet here, we’re talking about how numbers can start them. How do you strike that balance?

SB – There is a road that numbers finish and a road that numbers start. If you want to know how many people receive their health care through employer-sponsored plans, numbers finish the discussion. If you want to have a discussion about deductibility of employee health insurance, etc., then I would say that numbers start the policy discussion. Hopefully, people grounded in the same numbers and same facts start the policy discussion, and if someone wants to say what the number is, then there is an answer to that question. So numbers answer some questions and help people support an informed debate about what to do next.

HF – And presumably constrain that debate by ensuring that it doesn’t stray off into the crazy.

SB – Yes. It constrains in the sense that you’ve got to relate your argument to that which has really happened. A lot of the speculation going forward either denies the past or looks at it out of context, or just doesn’t mention it. There’s a very good journalist talking about the health-care bill debate — if you try to look historically, usually people don’t put it in a historical context. In fact the CMS data in some of these areas is two or more years old, which makes it hard to evaluate some of the impacts of the ACA, for example.

HF – So government tries to keep its data consistent from year to year, but problems creep in, there are changes in definitions and so on. Do you try to make sure that people using the data are aware of some of the problems they can run into?

SB – We do try to give a little bit of commentary. And mostly what we do is that if the data changed, we ignore the data before the point of change. We’re not trying to make this for rocket scientists. Take the SPM — the supplemental poverty measure. It didn’t exist before a certain period of time. We don’t try to synthesize, or talk about what it might have been. We just report it. Some data runs out — people stop measuring something. We can bemoan it, but we’re stopping where it stops, because we have a historical perspective going back perhaps to 1980. I can say we address it in that sense.

HF – Do you have ongoing conversations with people in the various statistical parts of the U.S. government about how best to use their data?

SB – Somehow, we and a staff member in the Treasury Department named Christina Ho found each other. They were working on something like an annual report, which was great. We were able to compare notes, and talk about data sets. Other than that, not yet. We just try to use the data from government as it’s provided. We do sometimes find that government data conflicts with itself, and in that case we have to find a way to highlight those conflicts, and hopefully push for some effort in government. I don’t know how anyone can make a decision when CMS data is contradictory as it relates to health care beneficiaries and the like. I do think that there has to be better work cross-government to reconcile different ways of sourcing data.

If a business is going to do things sensibly, it has to reconcile its data. The SEC [Securities and Exchange Commission] requires that businesses reconcile their data. Auditors insist that businesses reconcile their data. Our government should reconcile its data.

HF – Auditors reconcile their data, but there are different international accounting standards for doing this – might there be similar problems [of inconsistency] for government?

SB – I don’t think so. But it would be great if some accounting standard were used, and it was written down and explained. Some parts of government use GAAP [an internationally recognized set of accounting rules], others use cash accounting, some do certain accruals and not others. Write it down, do it consistently across the government and we can live with anything. Do I believe it would be wise to use GAAP in the U.S.? Sure. There are other international accounting standards but if we’re talking about something primarily for U.S. citizens, GAAP would be a reasonable one to use.

There are real issues. For example, do you book a liability for Social Security and Medicare on the balance sheet? You could put a pension liability on for employees, but essentially government runs a pension plan for the citizens. You could say that’s a forecast, but it’s one place where businesses and their 10-Ks [reporting forms required by the SEC] are forced to make a forecast. Whenever you have to approve pension liabilities, that is the kind of question you have to ask if you want to go to accrual accounting.

This article is one in a series supported by the MacArthur Foundation Research Network on Opening Governance that seeks to work collaboratively to increase our understanding of how to design more effective and legitimate democratic institutions using new technologies and new methods. Neither the MacArthur Foundation nor the Network is responsible for the article’s specific content. Other posts in the series can be found here.