Paul Baran (born 1926) was one of the developers of packet-switched networks along with Donald Davies and Leonard Kleinrock. He was born in Poland, but his family moved to Boston in 1928. Baran did undergraduate work at Drexel University, obtained his Masters degree in Engineering from UCLA in 1959 and began working for the RAND Corporation in the same year. The development of a communication network that would withstand a nuclear attack was important to US defence strategy. Baran developed his ideas for packet-switched networks as a solution. Similar ideas were also being independently pursued by Donald Davies from the National Physical Laboratory in the UK and Leonard Kleinrock at MIT. Baran also provided a spark of invention to four other important networking technologies. He was involved in the origin of the packet voice technology developed by StrataCom at its predecessor, Packet Technologies. This technology led to the first commercial pre-standard ATM product. He was also involved with the discrete multitone modem technology developed by Telebit, which was one of the roots of Orthogonal frequency-division multiplexing which is used in DSL modems. Paul also founded Metricom, the first wireless Internet company which deployed the first public wireless mesh networking system and Com21 an early cable modem company. In all cases, he provided early ideas and gave credibility to strong groups of developers who then took those ideas far beyond Baran's original spark. Paul Baran also extended his work in packet switching to wireless-spectrum theory, developing what he called "kindergarten rules" for the use of wireless spectrum. In addition to his innovation in networking products, he is also credited with inventing the metal detector used in airports.

IP addresses are unique on a given network. Networks are divided into subnets. A Subnet is specified by the network mask. If you look at an ip address and network mask in binary (as machines do) it all makes more sense. A common network mask you will see in a private ip network is (in binary)

11111111 11111111 11111111 00000000

grouped 8 bits at a time for clarity

That is 255.255.255.0 look familiar?

The mask tells us that any address that is to be considered part of our network must match exactly wherever there is a binary 1 and can differ anywhere there is a binary 0. Thus the addresses:

11000000 10101000 00000000 00000001 (192.168.0.1)

11000000 10101000 00000000 00000010 (192.168.0.2)

11000000 10101000 00000000 00000011 (192.168.0.3)

are the same when we look in the columns that have binary 1 in the mask (ie the first 3 decimal numbers) and differ only in the columns where there is a binary 0.

An address like

11000000 10101000 00000001 00000010 (192.168.1.2)

differs in a column where there is a binary 1, so this address is not in our subnet.

Thus, to actually answer the question, the parts of the address that are the same in different computers (if they are on the same subnet) is determined by the network mask.