It has not always been public. Most of it is now public, but there was a period (mostly before "Climategate") in which the relevant info absolutely most definitely was not public. There were reasons given why it was not public, but not-public it remained, and FOIA efforts (though not so much this particular one) did finally shake it loose.

[This is all from memory so I'm sure some of it is wrong, but If I Recall Correctly...]

This is practically ancient history now, but prior to the 1990s some of the best raw temperature data was collected and assembled and considered owned by individual national bureaucracies some of whom were sensitive about sharing it - they wanted to stay in the loop about how "their" data was used. To add all those national records together and make a worldwide temperature record, it was necessary to talk to all these bureaucrats and assuage their concerns. Thus, NASA and Hadley CRU apparently collected some of this info under legal agreements that said they wouldn't just let the rest of the world have the data.

So when NASA made GISS and Hadley made HADCRUT, they took the raw data, applied various corrective factors internally, and released a composite already-corrected data product. The outside world could see the adjusted climate data but could NOT view the "raw" data and QA it themselves or apply corrections and QA the effect of the corrections themselves.

This lack of transparency was a problem because periodically NASA would decide they should fix some problem, tweak the algorithm, and release a new temperature chart that had CHANGED, and there was no way for outside scientists to verify what they'd done and that they'd done it correctly - it was a black box.

Once the IPCC really got rolling and climate started to seem like A Big Deal, people like Steve McIntyre (at ClimateAudit) wanted to sanity-check what NASA and CRU were doing. They wanted to look at the raw, uncorrected data and at the algorithms used to clean it up. So they filed FOIA requests to ask for this data.

But (according to NASA at the time) the raw data was in a legal limbo - they had collected some of it under do-not-redistribute agreements. You couldn't release the WHOLE data set because it's divisible into parts that are restricted. No problem, says the ClimateAudit crowd, we'll just file separate requests - one per country - so you can give us the bits of data you CAN legally release and we'll work around the problem. Clever, right?

But (it now appears) nobody had carefully kept track of WHICH countries' data was restricted in that way. So it couldn't be done. The upshot was that our fine national scientists couldn't release ANY of the raw data without somebody first going back to ALL the national governments and getting their permission to do so.

Which was eventually (finally) done. And now the stuff is mostly aboveboard.

Alas, a lot of the "skeptic" attitude was generated during the keeping-lots-of-secrets era. There was a popular attitude that if scientists are hiding something it must be because they're doing something nefarious. Me, on the other hand, I tend to ascribe most of the secretiveness of that era to greed, incompetence, and embarrassment more than malice.

The "greed" part of it is that if you call up, say, Nigeria, and ask for a plot of their temperature record for the last year, there is some specific person whose job it was to massage that data and do stuff with it using some ancient computer and if he let the data go then some bright 14-year-old could do the same stuff on a cheap linux machine, rendering the bureaucrat's job obsolete. There were people who justified their budget on the basis of "we publish this Very Important Time Series" - it might not be the most important thing they do but it's the most VISIBLE thing they do - where publishing that time series was once really hard but by the 1990s is something any kid with a shell script could do in a half hour for free if data and algorithms were fully available.

The "embarrassment/incompetence" part of it is that if any of these scientists made some dumb mistake - maybe accidentally overwrote some of the data or had a bug in their corrections - making everything public could reveal this. So it has a possible downside and no obvious upside.

(Some climate data is still being kept private for these kinds of reasons, but mostly not by governments anymore. The main problematic category that still exists is data held by individual researchers. People who collect stuff like tree rings or sediment cores still will publish papers based on data they've never publicly archived - they do this because the data has value to them - they can publish more papers themselves or demand credit on other people's papers without fear of getting scooped. Most journals now require archiving data as a condition of publication but this requirement is widely ignored, delayed, or only minimally abided by.)