I just realized that the directions in "UPload" and "DOWNload" seem arbitrary to me as a non-native English speaker. I took a look at a couple of dictionaries and they said that this word is a result of merging "down" and "load", which doesn't seem to explain anything. Where could those two directions come from?
Initially, "download" and "upload" were used in aviation, especially by the US military. "Download" meant to remove items such as weapons from the aircraft, while "upload" meant to load items onto the aircraft.
For example, the August 1963 Aerospace Maintenance Safety (a publication of the US Air Force) says at page 18:
Failure to follow written procedures and download the missiles...
(meaning failure to remove the missiles from the aircraft)
There are earlier examples of "download", "downloading" and "uploading" in the January 1961 Aerospace Accident and Maintenance Review, also a USAF publication.
And still earlier in the October 1959 Aircraft Accident and Maintenance Review, USAF, at page 27:
During downloading of armament for a routine check, it was discovered that all the missiles aboard the F-102 had their internal power source activated.
(There was also an even earlier meaning relating to the direction of load on an aircraft component, such as on the tail of the aircraft. See "download on tail" in the April 1957 NACA Technical Note 3961 and ""download applied to the horizontal tail surface" in the 1952 US Code of Federal Regulations ).
Then, within the US Air Force, the concept was extended to computers.
The July 1968 IMPLEMENTATION OF THE USAF STANDARD BASE SUPPLY SYSTEM: A QUANTITATIVE STUDY says:
ADC provided a three-man team, which visited the bases some 30 days prior to conversion and conducted a full-scale download of the 305 and upload of the 1050, requiring 10 to 15 days.
- Download records from the previous computer
- Upload records on the 1050 computer
OED noticed the word for the first time in 1976, it seems.
1976 was very early in the development of computers. FTP (File Transfer Protocol) was designed 1971 and TCP/ IP followed in 1973. These two protocols made any transfer of data possible for the first time.
Servers or internet wasn't something anyone could imagine yet. Instead scientist were still hammering out the basic how of a data transfer.
In these years, several conceptual models for the transfer of data were developed. Most used and known nowadays is the OSI model with 7 layers - and these layers have an up / down direction.
Now, if I want to get data from another computer, I technically send the other computer a request, which then converts said data "down" to the physical layer, until all is left is a physical signal. This signal is then send to my computer, and I convert it again (this time "upwards").
This also is a likely explanation why "upload" appeared years later, in 1980 according to OED. Upload was coined as the simple opposite of download.
Though, while this is what I first thought of, it might be that there is also a hierarchy meaning in play as well.
Looking at the OED first use, "1976 Science 7 May 518/1 Software at any level can be developed on a host minicomputer and ‘down-loaded’ without code conversion", it appears to be talking about loading code from a host to a microprocessor. No real help there but I wonder if the author was thinking of taking reference books down in a library to copy out texts and then putting them back up on the shelves?
Download and Upload are far from arbitrary, they are entirely rational:
A Download refers to data that is brought 'down' from a network, the World Wide Web (Or Cloud) to reside on a local drive / computer.
Upload is data that is sent 'up' to the World Wide Web (Or Cloud) from a local drive of computer.
The term originated as others have described from the pre-personal computer era, and is a Networking term.
In the good o'l days [1970's],all computing power was housed exclusively in remote Server or mainframe computers - it was too expensive for an individual to buy.
Other than a local terminal directly wired to the 'glowing mass' of the mainframe, any other connections were "Down the wire", outside of a large company with a LAN, typically this would be the public telephone system. I remember visiting my Secondary School during the summer holidays in 1977 to play "Lunar Lander" via the dial-up telephone-cradle modem link to the local council mainframe computer.
As an aside: The game required the player to type in a numerical value, which was sent 'up' the telephone line to the mainframe, which would respond to the teleprinter with a message like:
"You are at 3,000' and descending at 50m/sec."
The player would type another value and so on until you received the message:
"Congratulations, you have crashed!"
Tv monitors and video displays were but a distant dream...
All local drives were "down the wire".
Any keyboard inputs were sent "up the wire"...
See also: Downstream