I have a project to geocode (i.e. use an API to give an address, get back latitude and longitude) many many addresses. Sometime, OSM works just fine, and provides just the right answer. It appears that 2/3 of the time this is true. 1/3 of the time, while OSM street data is aware enough to understand street names, it has no way of understanding the numbers. All of the sophisticated OSM guis have a mechanism to add this information. But this is very very complex to use. You have to be able to understand the "address range math" that OSM supports. While the math is very smart, the interface to it is very confusing. The current GUI too does a wonderful job of making tracing and "theres a thing of type X here" very simple, but the ability to add addressing is pretty deep, hidden and confusing. I would like to be able to contribute in some kind of crowd sourced fashion, where I tell my users "hey you might know where this single address is... help me out and enter it into this website!!" With enough individual address samples, the OSM algorithms (I assume) can figure out an entire area.. but I see no such interface. The website is inscrutable to outsiders... Would some OSM person please reply with a "you should try X?" Thanks, -FT

I would like to respond generally to some of the high ranked comments.

First, one of the assumptions is that an EHR is -one- kind of thing and it needs to be certified. This is much more a category buster, like a car. If a required car certification mandated that all cars should have beds like a truck, be able to off-road and break 150 mph, then you would have a tremendous change in how the auto-industry works. Even if you have seemingly reasonable requirements like "auto-door locks" or "automatic transmission". Not everyone -wants- a car like that.

To further complicate the problem, clinical clients typically have no idea what they want or need. Dr. Valdes (of LinuxMedNews is fond of saying "doctors have no idea what they want and programmers give it to them"

Also an important thing to recognize is that CCHIT is not primarily targeted at hospital systems, but "ambulatory" clinics. Hospitals are much smarter buyers and so certification has less power there. In the "ambulatory" market, the certification is taken as a short-cut for "good". If the certification is mandated you have a real problem

The other thing is that cost is only one part of the reason that the standard is broken. It is fundamentally incompatible with the FOSS model.

Consider a large company, like IBM that starts to sell support for a FOSS EHR project (like they did with GNU/Linux). Lets imagine that the supported OpenEMR.

OpenEMR is made up of 5-15 small companies (often one man shops) that have developed a pretty impressive EHR, given their resources. If IBM decide to support the codebase however, IBM would suddenly have 100s of clients, but would have contributed nothing to the actual development, which has been going on for close to a decade.

Then IBM takes OpenEMR to get certified. Because they have 100 clients that is a justifiable cost for them, they can split the cost between their clients.

Now, Rod Roark over at SunsetSystems is a core OpenEMR developer and might have 10 clients. (who knows how many he actually has of course)

So does IBM's CCHIT certified OpenEMR allow Rod to advertise his code as certified?

If the answer is "yes" then IBM has footed the bill for Rod's certification (This is essentially what WorldVistA did, but remember, they are a non-profit). IBM would not be happy about that.

If the answer is "no" then Rod cannot compete with IBM on a codebase that he has contributed far more to than IBM. OpenEMR is 'certifiable' because of Rod's work, but he has no benefit yet. Instead he has to go and spend $100k -again- to get -his- version of the codebase certified. But he does not have the clients to support the certification cost, even if he were able to pass all of the tests.

So I am not saying I want a "handout" for FOSS systems. But as you can see, the whole philosophy of CCHIT certification is designed to work with a company and a proprietary 'product'. Rather than a community of people and companies and a 'project'.

Some of your comments seem to reflect this, but others do not. I hope this makes my position clearer.

Thanks for reading!!

Fred Trotter