Converge Digest · Industry memory

The Technology Graveyard

The technologies that wired the world — and were replaced by it. Dead, dying, undead and a rare few resurrected, they opened the doors to the networks we depend on today. Every standard here solved a real problem, and the best of their ideas live on in whatever replaced them. Up from the ashes grow the roses of success. This archive honors that work.

198technologies remembered
159laid to rest
4resurrected
1876–2026years covered

198 of 198

The Technology Graveyard Quiz

Do you remember what replaced it?

Ten from the archive. Each one solved a real problem in its day — see how many successors you can still name.

Networking (Core)

CDMA / EV-DO (cdma2000)

1995 – 2022 Wireless (2G/3G)

How it workedEvery call was spread across a wide band by a unique pseudo-random code; receivers correlated against that code to extract their signal while all users shared the same spectrum at once.

Status
Dead
Killer / successor
LTE
Standards body
3GPP2 / TIA

2G GSM / 3G UMTS-HSPA

1991 – 2020 Wireless (2G/3G)

How it workedGSM used TDMA time-slots with SIM-based identity; UMTS moved to 5 MHz wideband CDMA, and HSPA added shared high-speed channels for real mobile data.

Status
Dying
Killer / successor
4G / 5G
Standards body
ETSI / 3GPP

FTP

1971 – 2020 File transfer

How it workedUsed separate control and data TCP connections to transfer files; the dual-channel design and cleartext credentials made it NAT-hostile and insecure.

Status
Dying
Killer / successor
HTTPS / SFTP
Standards body
IETF

Intel Omni-Path

2015 – 2020 HPC/AI fabric

How it workedA 100 Gbit/s HPC fabric with fine-grained link-layer flow control and packet-level error detection, designed to scale to very large node counts as an InfiniBand alternative.

Status
Resurrected
Killer / successor
(Rescued) - now pitched vs. InfiniBand
Standards body
Intel proprietary

SONET / SDH

1988 – 2020 Optical transport

How it workedA synchronous TDM hierarchy that framed tributaries into standard optical rates (OC-3/12/48...), multiplexed by a master clock, with fast (<50 ms) ring protection over redundant fiber paths.

Status
Dying
Killer / successor
OTN, packet-optical, Ethernet
Standards body
ANSI T1X1 / ITU-T

ISDN

1986 – 2018 WAN / access

How it workedDigitized the local loop into 64 kbit/s B-channels for voice/data plus a 16 kbit/s D-channel for signaling (BRI = 2B+D), giving all-digital dial-up and faster call setup over ordinary copper.

Status
Undead
Killer / successor
DSL/broadband; VoIP
Standards body
ITU-T

WiMAX (802.16)

2001 – 2015 Wireless broadband

How it workedOFDMA-based fixed/mobile broadband: base stations scheduled uplink/downlink slots to subscriber stations across wide channels with carrier-grade QoS - on the losing side of the LTE ecosystem.

Status
Dead
Killer / successor
LTE
Standards body
IEEE 802.16; WiMAX Forum

iDEN

1994 – 2013 Wireless (PTT)

How it workedMotorola's TDMA system that layered digital cellular voice, data, and near-instant push-to-talk (Direct Connect) onto fragmented SMR radio channels.

Status
Dead
Killer / successor
LTE / CDMA
Standards body
Motorola proprietary

IPv4

1981 – 2011 Internet protocol

How it worked32-bit addressing with best-effort datagram delivery and (post-CIDR) classless routing; NAT stretched its ~4.3 billion addresses far past exhaustion, keeping it alive despite IPv6.

Status
Undead
Killer / successor
IPv6
Standards body
IETF

BPL (Broadband over Power Line)

2004 – 2010 Access

How it workedInjected high-frequency data signals onto medium/low-voltage power lines, using the electrical grid as the last mile - but radiated interference and noisy lines undid it.

Status
Dead
Killer / successor
DSL / cable / fiber
Standards body
HomePlug / IEEE 1901

FCoE

2009 – 2010 Data center fabric

How it workedEncapsulated Fibre Channel frames directly inside Ethernet (requiring lossless Data Center Bridging enhancements) so SAN and LAN traffic could share one converged 10 GbE adapter and fabric.

Status
Undead
Killer / successor
iSCSI; native FC; NVMe-oF
Standards body
INCITS T11

Frame Relay

1990 – 2010 WAN / carrier

How it workedSwitched variable-length frames over pre-provisioned permanent virtual circuits, dropping X.25's per-hop error correction (links were now clean) and pushing recovery to the endpoints for far higher throughput.

Status
Undead
Killer / successor
MPLS / IP-VPN
Standards body
ITU-T / ANSI; Frame Relay Forum

HomePNA (phoneline)

1998 – 2010 Home networking

How it workedNetworked PCs over existing in-home telephone wiring by using frequencies above the voice band, avoiding new cabling.

Status
Dead
Killer / successor
Wi-Fi / MoCA / Ethernet
Standards body
HomePNA Alliance

NNTP / Usenet

1979 – 2010 Messaging

How it workedPropagated threaded newsgroup articles server-to-server by flood-fill store-and-forward, so every participating server eventually held a copy of each post.

Status
Dying
Killer / successor
Web forums; social media
Standards body
IETF

Wireless USB (UWB)

2005 – 2010 PAN

How it workedCarried USB over ultrawideband radio (3.1-10.6 GHz) at short range, spreading signals across a huge bandwidth at low power for cable-free peripherals.

Status
Dead
Killer / successor
Wi-Fi / Bluetooth
Standards body
USB-IF / WiMedia

AppleTalk / LocalTalk

1985 – 2009 LAN / protocol

How it workedA plug-and-play stack where nodes self-assigned addresses and discovered services by name (NBP); LocalTalk ran it at 230 kbit/s over daisy-chained shielded cable.

Status
Dead
Killer / successor
TCP/IP
Standards body
Apple proprietary

Quadrics QsNet

1996 – 2009 HPC interconnect

How it workedA fat-tree HPC network with a programmable NIC that offloaded remote-memory (RDMA-style) operations, giving very low latency plus hardware barrier and collective support for supercomputers.

Status
Dead
Killer / successor
InfiniBand
Standards body
Proprietary

Banyan VINES

1984 – 2003 LAN / NOS

How it workedA Unix-based network OS whose StreetTalk global directory let users find resources across a wide-area internetwork by name, ahead of its time but tied to VINES' own protocol.

Status
Dead
Killer / successor
TCP/IP; NetWare/NT
Standards body
Banyan proprietary

NetBEUI

1985 – 2001 LAN / protocol

How it workedA lightweight, non-routable LAN transport that carried NetBIOS names by broadcast within a single segment - fast for small workgroups but unable to cross routers.

Status
Dead
Killer / successor
TCP/IP
Standards body
IBM / Microsoft

APON / BPON

1998 – 2000 Access (PON)

How it workedEarly passive optical networks: one fiber from the central office fanned out through unpowered splitters to homes, sharing ATM-cell (APON) or broadband (BPON) capacity by time-division.

Status
Dead
Killer / successor
GPON / XGS-PON
Standards body
ITU-T / FSAN

ATM (Asynchronous Transfer Mode)

1988 – 2000 WAN / carrier

How it workedChopped all traffic into fixed 53-byte cells (48 payload + 5 header) switched over virtual circuits, so hardware could switch at constant speed and offer per-connection QoS for voice, video and data on one network.

Status
Undead
Killer / successor
IP / Ethernet
Standards body
ITU-T; ATM Forum

DECnet

1975 – 2000 Protocol stack

How it workedDigital's peer-to-peer networking stack (Phase IV/V) linking VAX/VMS systems with area-based routing, once a major enterprise internetwork before TCP/IP.

Status
Dead
Killer / successor
TCP/IP
Standards body
DEC proprietary

Dial-up / analog modems

1962 – 2000 WAN / access

How it workedModulated digital data onto audible carrier tones within the 300-3400 Hz voice band (QAM/trellis coding), negotiating the fastest mutually supported rate over a normal phone call.

Status
Dead
Killer / successor
Broadband (DSL/cable)
Standards body
ITU-T (V-series)

ESCON

1990 – 2000 Mainframe I/O

How it workedIBM's fiber channel-to-control-unit link that replaced bulky parallel 'bus and tag' copper with serial optical connections (~17 MB/s) switched through ESCON directors.

Status
Dead
Killer / successor
FICON (FC-based)
Standards body
IBM

IGRP

1985 – 2000 Routing protocol

How it workedCisco's distance-vector routing protocol using a composite metric (bandwidth, delay, load, reliability) instead of hop count, later superseded by its own classless EIGRP.

Status
Dead
Killer / successor
EIGRP / OSPF
Standards body
Cisco proprietary

InfiniBand

2000 – 2000 HPC/AI fabric

How it workedA switched, RDMA-native fabric where host channel adapters move data directly between nodes' memory with OS-bypass and credit-based lossless flow control - hence its dominance in AI/HPC clusters.

Status
Resurrected
Killer / successor
(Reborn) - the AI-fabric default
Standards body
IBTA

IPX/SPX (NetWare)

1983 – 2000 LAN / protocol

How it workedNovell's routable protocols derived from Xerox XNS; IPX carried connectionless datagrams keyed to network+node addresses, with SAP broadcasts advertising file and print services.

Status
Dead
Killer / successor
TCP/IP
Standards body
Novell proprietary

IrDA (infrared)

1993 – 2000 PAN

How it workedShort-range, line-of-sight data links using modulated infrared LEDs between devices (up to 4 Mbit/s), needing direct aim and close proximity.

Status
Dead
Killer / successor
Bluetooth
Standards body
IrDA

LMDS / MMDS

1998 – 2000 Fixed wireless

How it workedFixed broadband over microwave: rooftop antennas linked to line-of-sight base stations in the 2 GHz (MMDS) or 28/31 GHz (LMDS) bands, but rain fade and line-of-sight limits capped reach.

Status
Dead
Killer / successor
Cable/DSL/fiber; later fixed LTE
Standards body
FCC / IEEE

Myrinet

1995 – 2000 HPC interconnect

How it workedA lightweight HPC interconnect using source-routed wormhole switching and OS-bypass messaging, delivering microsecond latency between cluster nodes far below Ethernet's.

Status
Dead
Killer / successor
InfiniBand / Ethernet
Standards body
Proprietary (later ANSI)

RSVP / IntServ

1997 – 2000 QoS

How it workedReserved bandwidth per data flow end-to-end: each router held state for every flow and admitted or denied it - precise QoS that collapsed under the state explosion in the core.

Status
Dead
Killer / successor
DiffServ; MPLS-TE
Standards body
IETF

SCI (Scalable Coherent Interface)

1992 – 2000 Interconnect

How it workedProvided cache-coherent shared memory across nodes over point-to-point rings, letting a cluster act like a single large NUMA machine.

Status
Dead
Killer / successor
PCIe / InfiniBand
Standards body
IEEE 1596

Telnet

1969 – 2000 Remote access

How it workedOpened a raw TCP session that relayed keystrokes and terminal output verbatim between client and host - simple, but everything including passwords travelled in cleartext.

Status
Dead
Killer / successor
SSH
Standards body
IETF

Token Ring (802.5)

1985 – 2000 LAN

How it workedStations passed a small 'token' frame around a logical ring; only the token-holder could transmit, giving deterministic, collision-free access at the cost of ring-management overhead.

Status
Dead
Killer / successor
Ethernet
Standards body
IEEE 802.5

WAP (Wireless Application Protocol)

1999 – 2000 Mobile data

How it workedDelivered stripped-down web content to feature phones using WML over a binary, gateway-optimized protocol stack tuned for slow, high-latency 2G links and tiny screens.

Status
Dead
Killer / successor
Real mobile HTML (post-iPhone)
Standards body
WAP Forum / OMA

X.25

1976 – 2000 WAN / carrier

How it workedCarried data in numbered packets over virtual circuits, with error-checking and flow control at every hop because the underlying analog links were noisy; nodes stored and retransmitted to guarantee delivery.

Status
Dead
Killer / successor
Frame Relay, then IP
Standards body
CCITT (now ITU-T)

100VG-AnyLAN

1995 – 1990 LAN

How it workedReplaced CSMA/CD with 'demand priority,' where a central hub polled ports round-robin and granted transmission slots, carrying Ethernet or Token Ring frames at 100 Mbit/s over four-pair copper.

Status
Dead
Killer / successor
Fast Ethernet (802.3u)
Standards body
IEEE 802.12

10BASE5 / 10BASE2 (coax)

1980 – 1990 LAN

How it workedShared half-duplex Ethernet on a single coaxial cable; stations tapped the bus (vampire taps or BNC tees) and used CSMA/CD to arbitrate one collision domain.

Status
Dead
Killer / successor
10BASE-T twisted pair
Standards body
IEEE 802.3

ARCNET

1977 – 1990 LAN

How it workedA token-passing bus/star LAN where a token circulated by station address (not physical position), giving predictable performance over coax at 2.5 Mbit/s.

Status
Dead
Killer / successor
Ethernet
Standards body
Proprietary (later ANSI 878.1)

FDDI

1987 – 1990 LAN / campus

How it workedA 100 Mbit/s dual counter-rotating fiber token ring; the second ring wrapped around a break to self-heal, and timed token rotation bounded latency for campus backbones.

Status
Dead
Killer / successor
Fast / Gigabit Ethernet
Standards body
ANSI X3T9.5

Gopher

1991 – 1990 Info service

How it workedServed hierarchical menus of documents and links over a simple text protocol; you navigated nested menus rather than following inline hyperlinks.

Status
Dead
Killer / successor
World Wide Web
Standards body
Univ. of Minnesota (RFC 1436)

OSI protocol suite (CLNP/TP4)

1984 – 1990 Protocol stack

How it workedThe ISO stack: CLNP was the connectionless network layer (IP's rival) and TP4 the reliable transport, addressed by lengthy NSAP identifiers - technically complete but heavy and late.

Status
Dead
Killer / successor
TCP/IP
Standards body
ISO/IEC; CCITT

SMDS

1991 – 1990 WAN / carrier

How it workedA connectionless, cell-based metropolitan data service: customers sent datagrams with E.164-style addresses into the carrier cloud, which delivered them without setting up a circuit.

Status
Dead
Killer / successor
Frame Relay / ATM
Standards body
Bellcore

Token Bus (802.4)

1982 – 1990 LAN / industrial

How it workedFormed a logical token-passing ring over a physical coax bus, so factory-floor stations got deterministic, collision-free access (favored by GM's MAP initiative).

Status
Dead
Killer / successor
Ethernet
Standards body
IEEE 802.4

WAIS

1989 – 1990 Networked search

How it workedIndexed and searched full-text document collections over the Z39.50 protocol, returning ranked results - an early networked search engine before the web.

Status
Dead
Killer / successor
Web search engines
Standards body
Consortium (TMC, Apple, etc.)

ADSL / ADSL2+

1999 – Access

How it workedSplit the copper-loop spectrum into hundreds of DMT subcarriers, assigning most bandwidth downstream; reach and rate fell off sharply with loop length.

Status
Dying
Killer / successor
FTTH; DOCSIS
Standards body
ITU-T

Fibre Channel

1994 – Storage (SAN)

Legacy standard still in active enterprise service.

How it workedA purpose-built, low-latency serial SAN fabric: switches route frames between host bus adapters and storage by 24-bit fabric addresses, with buffer-credit flow control to avoid drops.

Status
Dying
Killer / successor
NVMe-oF over Ethernet
Standards body
INCITS T11

H.323

1996 – VoIP signaling

How it workedAn ITU umbrella of protocols for packet voice/video: gatekeepers handled admission and address resolution while H.245/RTP carried the media - comprehensive but heavyweight versus SIP.

Status
Dying
Killer / successor
SIP
Standards body
ITU-T

IP Multicast (interdomain)

1988 – Protocol

How it workedRouters built distribution trees (via IGMP/PIM) so one packet stream reached many receivers without duplication - elegant within a domain, but interdomain trust, addressing and billing never scaled.

Status
Undead
Killer / successor
CDNs; unicast
Standards body
IETF

MGCP / Megaco (H.248)

1999 – VoIP signaling

How it workedA master/slave model where a central call agent (softswitch) controlled dumb media gateways, instructing them how to set up and tear down media streams.

Status
Dying
Killer / successor
SIP
Standards body
IETF / ITU-T

Mobile IP

1996 – Protocol

How it workedGave a device a stable 'home' IP while roaming, tunneling traffic from a home agent to its current care-of address - an elegant scheme the mobile core ultimately solved differently (GTP).

Status
Undead
Killer / successor
GTP; PMIPv6 (mobile core)
Standards body
IETF

MPLS

1997 – WAN / carrier

Legacy standard still in active enterprise service.

How it workedRouters prepend a short label to each packet; downstream label-switched routers forward by label swap along pre-computed paths (LSPs) instead of per-hop IP lookups, enabling traffic engineering and VPNs.

Status
Dying
Killer / successor
SD-WAN
Standards body
IETF

PDH (T1/E1/DS3)

1962 – WAN / access

How it workedTime-division multiplexing that interleaved 24 (T1) or 30 (E1) 64 kbit/s voice channels into a frame, then stacked those into higher rates by bit-stuffing loosely synchronized streams.

Status
Dying
Killer / successor
SONET, then packet
Standards body
AT&T/Bell; ITU-T (E1)

POTS / PSTN

1876 – Telecom access

How it workedAnalog voice carried as a continuous electrical signal over a copper loop to a circuit-switched exchange, which set up an end-to-end dedicated path for the duration of the call.

Status
Dying
Killer / successor
VoIP; mobile
Standards body
ITU-T

SS7

1980 – Telecom signaling

How it workedOut-of-band signaling: call-control messages (setup, teardown, 800-number lookups) travelled on a separate packet-data channel from the voice trunks, speeding call setup and enabling intelligent-network features.

Status
Undead
Killer / successor
Diameter / SIP-I (all-IP core)
Standards body
ITU-T / ANSI

Zigbee / Z-Wave

2001 – IoT mesh

How it workedLow-power, low-rate mesh networking for sensors and home automation: nodes relay small packets hop-to-hop (Zigbee on 802.15.4/2.4 GHz; Z-Wave sub-GHz) to extend range on tiny batteries.

Status
Dying
Killer / successor
Thread / Matter
Standards body
Zigbee/CSA; Sigma/Z-Wave Alliance

Telecom / PSTN

TWX (Teletypewriter Exchange)

1931 – 2000 Data service

How it workedA switched network of teleprinters: subscribers dialed each other like phone calls but exchanged typed messages over the voice network at ~45-110 baud, printing on paper at both ends.

Status
Dead
Killer / successor
Fax, then email / IP
Key players
Bell Labs / Western Electric / AT&T

Step-by-Step (Strowger / SxS)

1919 – 1999 Switching

How it workedDirect control: each dialed digit pulsed a two-motion selector that stepped up then rotated to the next stage, wiring the call contact-by-contact as you dialed. No central brain - the dial literally drove the switch train.

Status
Dead
Killer / successor
Crossbar and ESS; digital by 1999
Key players
Bell Labs / Western Electric / AT&T; Almon Strowger

Crossbar (No. 1 / No. 5)

1938 – 1990 Switching

How it workedA grid of horizontal and vertical bars: energizing one bar on each axis closed exactly one crosspoint to build a path. Fast 'markers' computed the route in milliseconds, set it up, then released to handle the next call.

Status
Dead
Killer / successor
Electronic (1ESS), then digital switching
Key players
Bell Labs / Western Electric / AT&T

TD-2 microwave radio relay

1950 – 1990 Transmission

How it workedLine-of-sight microwave hops between horn-reflector antennas on towers ~30 miles apart relayed multiplexed long-distance voice and network TV without wires; each tower received, amplified, and re-beamed the signal to the next.

Status
Dead
Killer / successor
Fiber optics; digital radio
Key players
Bell Labs / Western Electric / AT&T

Magneto (local-battery) telephone

1890 – 1983 Subscriber set

How it workedNo power from the exchange: the handset spoke on its own dry cells, and turning the hand crank spun a magneto generator to push ~90 V AC ringing current down the line to the operator or party-line bells.

Status
Dead
Killer / successor
Common-battery / dial telephones
Key players
Bell Labs / Western Electric / AT&T

Manual switchboard (cord board)

1878 – 1983 Switching

How it workedOperators sat at a wall of jacks; a caller's signal dropped a metal flag, the operator plugged a cord pair between the calling and called jacks and rang the far end by hand. Every call was set up and torn down by a person.

Status
Dead
Killer / successor
Automatic switching (panel, SxS, crossbar)
Key players
Bell Labs / Western Electric / AT&T

Panel switch

1915 – 1983 Switching

How it workedBell's first big automatic switch. Dial pulses were captured by common-control 'senders,' which drove brushes climbing tall vertical panels of terminals by electric motor to hunt for an idle path - the customer dialed, machinery searched.

Status
Dead
Killer / successor
Crossbar; then ESS
Key players
Bell Labs / Western Electric / AT&T

L-carrier coaxial (analog FDM)

1941 – 1980 Transmission

How it workedAnalog frequency-division multiplexing stacked thousands of 4 kHz voice channels into one broadband spectrum on coaxial tubes, with buried repeater huts every few miles re-amplifying the signal coast to coast (L5 carried ~108,000 circuits).

Status
Dead
Killer / successor
Digital carrier and fiber optics
Key players
Bell Labs / Western Electric / AT&T

Open-wire lines

1880 – 1980 Transmission

How it workedBare copper conductors bolted to glass insulators on wooden crossarms; wires were periodically transposed (crossed over) to cancel crosstalk and interference. Weather-exposed and very low capacity.

Status
Dead
Killer / successor
Cable, carrier, then fiber
Key players
Bell Labs / Western Electric / AT&T

Party lines

1880 – 1980 Access

How it workedSeveral households shared one physical loop; coded ring cadences or frequency/grounded-ringer schemes signaled which subscriber was being called - and neighbors could listen in or tie up the line.

Status
Dead
Killer / successor
Dedicated single-subscriber loops
Key players
Bell Labs / Western Electric / AT&T

In-band SF / MF signaling

1920 – 1976 Signaling

How it workedTrunk control rode inside the voice channel as audible tones: a 2600 Hz tone idled or seized a trunk, and multifrequency (MF) tone pairs 'outpulsed' the dialed number. Because control shared the talk path, a matching tone from a handset could seize trunks - the blue-box exploit.

Status
Dead
Killer / successor
CCIS (SS6) / SS7 - out-of-band
Key players
Bell Labs / Western Electric / AT&T

Picturephone (Mod II)

1970 – 1973 Terminal / service

How it workedTwo-way video calling over specially conditioned lines needing ~1 MHz - about three voice circuits - each direction, with a small CRT and camera in one desktop unit. Priced far above the value delivered; Pittsburgh peaked at 32 sets in 1972.

Status
Dead
Killer / successor
(None then) - later IP video
Key players
Bell Labs / Western Electric / AT&T

Rotary dial pulse signaling

1919 – 1963 Signaling

How it workedThe rotary dial briefly broke the loop current a set number of times per digit (ten pulses = '0'); downstream switches counted the interruptions to read the number.

Status
Dying
Killer / successor
DTMF (Touch-Tone)
Key players
Bell Labs / Western Electric / AT&T

Wireless & Cellular

i-mode

1999 – 2026 Mobile internet (Japan)

How it workedThe world's first mass mobile-internet service: packet data on feature phones using compact HTML, a walled garden of carrier-billed content and services, and the original emoji set. Peaked near 49M users in 2010.

Status
Dead
Killer / successor
Smartphone web; 5G
Key players
NTT DoCoMo

PHS (Personal Handy-phone System)

1995 – 2023 Cordless-cellular

How it workedA hybrid of cordless and cellular: dense, low-power microcells - essentially public cordless base stations - gave cheap calls and data but handed off poorly at speed, so it dropped calls in moving cars.

Status
Dead
Killer / successor
3G / LTE
Key players
NTT, DDI Pocket, WILLCOM

cdmaOne (IS-95)

1995 – 2022 2G digital

How it workedQualcomm's code-division approach: every call was spread across a wide band with a unique code, so all users shared the same spectrum at once and the network sorted them by code, not by channel or time slot.

Status
Dead
Killer / successor
LTE
Key players
Qualcomm, Verizon, Sprint

GSM 2G / GPRS / EDGE

1991 – 2020 2G / 2.5G

How it workedTDMA digital voice with SIM-based identity; GPRS then EDGE bolted packet data onto the same channels to deliver the first 'always-on' mobile internet.

Status
Dying
Killer / successor
4G / 5G
Key players
ETSI, Ericsson, Nokia

One-Seg (1seg mobile TV)

2006 – 2020 Mobile TV (Japan)

How it workedOne 1/13th segment of Japan's ISDB-T digital TV signal was reserved to broadcast free live TV to phones and car navs, no cellular data needed.

Status
Dying
Killer / successor
Video streaming
Key players
NHK, Japanese broadcasters

TD-SCDMA

2009 – 2020 3G

How it workedChina's home-grown 3G using time-division duplex on unpaired spectrum, deployed almost solely by China Mobile under government mandate.

Status
Dead
Killer / successor
TD-LTE
Key players
China Mobile, Datang, Siemens

UMTS / W-CDMA

2001 – 2020 3G

How it worked3G wideband CDMA on new 5 MHz carriers, built for real mobile data and video calling.

Status
Dying
Killer / successor
LTE (4G)
Key players
3GPP, NTT DoCoMo, Ericsson

Pocket Bell / pagers (pokeberu)

1968 – 2019 Paging (Japan)

How it workedNumeric and text pagers that became a 1990s teen craze in Japan, with elaborate number-code slang; one-way (later limited two-way) messages over dedicated paging spectrum.

Status
Dead
Killer / successor
PHS / keitai; SMS
Key players
NTT, Tokyo Telemessage

WiBro

2006 – 2018 Mobile broadband (Korea)

How it workedKorea's deployment of mobile WiMAX (IEEE 802.16e) - an early, genuinely leading mobile-broadband network years before LTE was widespread.

Status
Dead
Killer / successor
LTE
Key players
KT, SK Telecom, Samsung

Xiaolingtong ('Little Smart')

1997 – 2014 Cordless-cellular (China)

How it workedA PHS-based service priced like a fixed line and marketed as extending your home number to a cheap handset via dense microcells - a way around China's expensive true-mobile tariffs. Peaked at 93M users in 2006.

Status
Dead
Killer / successor
3G (spectrum reclaimed)
Key players
China Telecom, China Unicom, UTStarcom

iDEN

1994 – 2013 2G digital (PTT)

How it workedMotorola's hybrid of TDMA cellular plus digital push-to-talk (Direct Connect walkie-talkie) running on fragmented SMR radio spectrum rather than cellular bands.

Status
Dead
Killer / successor
LTE
Key players
Motorola, Nextel

PDC (Personal Digital Cellular)

1993 – 2012 2G digital

How it workedJapan's home-grown 2G TDMA, deployed only in Japan and incompatible with everything else.

Status
Dead
Killer / successor
W-CDMA (3G)
Key players
NTT DoCoMo, Japanese vendors

DVB-H

2007 – 2011 Mobile TV (Europe)

How it workedThe European standard for broadcasting digital TV to battery-powered handsets, trialed widely across the EU.

Status
Dead
Killer / successor
Video streaming
Key players
Nokia, European broadcasters

MediaFLO

2007 – 2011 Mobile TV (US)

How it workedQualcomm's dedicated broadcast network transmitting live TV channels to phones on its own 700 MHz spectrum, separate from the cellular data network.

Status
Dead
Killer / successor
Video streaming over 3G/4G
Key players
Qualcomm

Mobitex / DataTAC (ARDIS)

1980 – 2010 Wireless data

How it workedNarrowband packet-radio networks carrying two-way pages and early BlackBerry email reliably years before cellular data existed - low speed, but robust and store-and-forward.

Status
Dead
Killer / successor
GPRS / 3G data
Key players
Ericsson, Motorola, RIM, IBM

ReFLEX two-way paging

1990 – 2010 Wireless data

How it workedPagers on dedicated high-power, low-frequency networks with a low-speed return channel for acknowledgments; the long wavelengths punched deep into buildings where early cell phones failed.

Status
Dying
Killer / successor
SMS / smartphones
Key players
Motorola, Glenayre, paging carriers

AMPS

1983 – 2008 1G analog

How it workedThe first true cellular system: analog FM voice with FDMA (one caller per 30 kHz channel) at 800 MHz, splitting a metro into cells that reused frequencies and handed calls off tower-to-tower as you drove.

Status
Dead
Killer / successor
Digital 2G (D-AMPS, GSM, CDMA)
Key players
AT&T / Bell Labs, Motorola

CDPD (Cellular Digital Packet Data)

1993 – 2008 Wireless data

How it workedSent IP packets over idle AMPS voice channels, hopping to whichever channel happened to be free between calls - data squeezed into the gaps of an analog voice network.

Status
Dead
Killer / successor
GPRS / 1xRTT
Key players
IBM, AT&T, carriers

D-AMPS / IS-136 (TDMA)

1993 – 2008 2G digital

How it workedOverlaid digital TDMA on the AMPS 30 kHz channels, slicing each into 3-6 time slots so several calls shared one frequency - roughly tripling capacity while staying backward-compatible with analog.

Status
Dead
Killer / successor
GSM / CDMA, then LTE
Key players
AT&T, Ericsson, Nokia

RTMS

1985 – 2005 1G analog (Italy)

How it workedItaly's home-grown analog cellular on 450 MHz, deployed only in Italy before the country adopted TACS and GSM.

Status
Dead
Killer / successor
TACS, then GSM
Key players
SIP / Telecom Italia

HomeRF

1998 – 2003 WLAN

How it workedA home wireless-LAN standard using frequency hopping at 2.4 GHz that blended cordless voice and data on one network.

Status
Dead
Killer / successor
Wi-Fi (802.11b)
Key players
Intel, Proxim, HP

Teledesic

1990 – 2002 Satellite

How it workedA planned 'Internet in the Sky' of hundreds of LEO satellites for global broadband - fully funded on paper, only a single test satellite ever flew.

Status
Dead
Killer / successor
(None then) - later Starlink
Key players
Bill Gates, Craig McCaw, Boeing

Iridium (original venture)

1998 – 2001 Satellite

How it worked66 cross-linked low-earth-orbit satellites relaying calls satellite-to-satellite so a handset could reach anywhere on Earth with no local infrastructure - a phone network in the sky.

Status
Resurrected
Killer / successor
(Rescued) - terrestrial cellular nearly killed it
Key players
Motorola

Metricom Ricochet

1994 – 2001 Wireless data

How it workedA mesh of pager-sized shoulder radios talking to pole-top microcell repeaters, delivering ~28-128 kbit/s wireless internet across a handful of cities.

Status
Dead
Killer / successor
Wi-Fi / 3G / 4G
Key players
Metricom (Paul Allen-backed)

TACS / ETACS

1985 – 2001 1G analog

How it workedAn AMPS-derived analog system adapted to 900 MHz for the UK, Europe and parts of Asia - same FDMA/FM approach on different channels.

Status
Dead
Killer / successor
GSM
Key players
Vodafone, Cellnet, Motorola

C-Netz (C450)

1985 – 2000 1G analog (Germany)

How it workedWest Germany's analog cellular on 450 MHz, notable for an early removable smartcard that identified the subscriber - a direct ancestor of the SIM.

Status
Dead
Killer / successor
GSM (D-Netz)
Key players
Deutsche Bundespost, Siemens

CT1 / CT2 cordless (Telepoint)

1980 – 2000 Cordless

How it workedEarly cordless-phone standards; CT2 even powered 'Telepoint,' a public scheme where you made outbound-only calls near street-corner base stations - a cordless payphone with no incoming calls.

Status
Dead
Killer / successor
DECT / cellular
Key players
European PTTs, GPT

ERMES

1990 – 2000 Paging (Europe)

How it workedA pan-European paging standard meant to unify the continent's fragmented national pager networks with roaming.

Status
Dead
Killer / successor
GSM / SMS
Key players
ETSI, European carriers

NMT (Nordic Mobile Telephone)

1981 – 2000 1G analog

How it workedNordic analog 1G with the first automatic international roaming - drive across a border and keep your call. Voice was analog FM on dedicated channels.

Status
Dead
Killer / successor
GSM
Key players
Nordic PTTs, Ericsson, Nokia

Radiocom 2000

1985 – 2000 1G analog (France)

How it workedFrance's national analog cellular system, incompatible with its neighbors' 1G systems.

Status
Dead
Killer / successor
GSM
Key players
France Telecom

ICO Global

1990 – 1999 Satellite (global/regional)

How it workedA planned medium-earth-orbit constellation for global satellite phones, an Iridium/Globalstar rival that filed for bankruptcy before commercial service.

Status
Dead
Killer / successor
(None) - terrestrial cellular
Key players
Inmarsat spinoff, Craig McCaw

NTT 1G (Automobile Telephone)

1979 – 1999 1G analog (Japan)

How it workedThe world's first commercial cellular network - analog car phones on a large-zone cell system in Tokyo, live three-plus years before AMPS. Voice was analog FM with automated handoff between big zones.

Status
Dead
Killer / successor
PDC (2G digital)
Key players
NTT, Matsushita, NEC

Bi-Bop (CT2 Telepoint)

1991 – 1997 Cordless / Telepoint (France)

How it workedA CT2 'Telepoint' service: you made outbound-only calls when standing near a marked street base station - a public cordless phone with no ability to receive calls.

Status
Dead
Killer / successor
GSM cellular
Key players
France Telecom

Rabbit (CT2 Telepoint)

1992 – 1993 Cordless / Telepoint (UK)

How it workedThe UK's CT2 Telepoint: outbound-only cordless calls near base stations in shops and stations. It launched into the teeth of real cellular and lasted barely a year.

Status
Dead
Killer / successor
GSM cellular
Key players
Hutchison

IMTS (Improved Mobile Telephone Service)

1964 – 1980 Pre-cellular

How it workedAdded automatic direct dialing and full-duplex over a single powerful transmitter per city - but still one giant 'cell,' so a metro had only ~10-40 channels and multi-year waiting lists.

Status
Dead
Killer / successor
AMPS cellular
Key players
AT&T / Bell Labs

MTS (Mobile Telephone Service)

1946 – 1960 Pre-cellular

How it workedHalf-duplex push-to-talk radiotelephone: a live operator patched the car phone into the wired network, and everyone in a city shared a few high-power channels, so you waited for a free one and couldn't talk and listen at once.

Status
Dead
Killer / successor
IMTS
Key players
AT&T / Bell Labs, Motorola

Optical

Resilient Packet Ring (RPR)

2004 – 2010 Metro transport

How it workedA dual counter-rotating fiber ring with a fairness MAC that shared bandwidth statistically and healed in under 50 ms - SONET-grade resilience with packet efficiency.

Status
Dead
Killer / successor
Carrier Ethernet / MPLS
Key players
Cisco, Nortel, IEEE

All-optical MEMS cross-connect

2000 – 2002 Optical switching

How it workedArrays of tilting micromirrors (MEMS) physically steered whole wavelengths between fibers with no conversion to electricity, promising an all-optical switching fabric.

Status
Dead
Killer / successor
Telecom crash; OEO economics
Key players
Lucent (LambdaRouter), Xros/Nortel, Calient

Cisco Dynamic Packet Transport (DPT/SRP)

1999 – 2000 Metro transport

How it workedCisco's proprietary Spatial Reuse Protocol ring - the precursor that seeded the later RPR standard.

Status
Dead
Killer / successor
RPR standard, then Ethernet
Key players
Cisco

OEO per-span regeneration

1980 – 1995 Long-haul transmission

How it workedEarly long-haul converted light back to electricity and re-transmitted it at every span to clean the signal - one regenerator per wavelength, per hut.

Status
Dead
Killer / successor
EDFA optical amplifiers
Key players
Long-haul carriers

CDDI (Copper FDDI)

1994 – 1990 LAN

How it workedRan the 100 Mbit/s FDDI token-ring protocol over twisted-pair copper instead of fiber, to cut cost to the desktop.

Status
Dead
Killer / successor
Fast Ethernet
Key players
ANSI, Crescendo/Cisco

FDDI-II

1990 – 1990 LAN

How it workedAdded isochronous, circuit-switched channels to FDDI so it could carry voice and video alongside data.

Status
Dead
Killer / successor
(Unused) - Fast Ethernet
Key players
ANSI X3T9.5

Ethernet over SONET (GFP/VCAT/LCAS)

2002 – Transport adaptation

How it workedGFP wrapped Ethernet into SONET, VCAT bonded arbitrary numbers of SONET timeslots, and LCAS resized that bundle hitlessly - bolting data neatly onto a voice-era TDM network.

Status
Dead
Killer / successor
Native packet / OTN
Key players
ITU-T, transport vendors

Optical Burst Switching (OBS)

2000 – Optical switching (research)

How it workedSent a control packet ahead to reserve an all-optical path, then fired a data 'burst' through without buffering it anywhere.

Status
Dead
Killer / successor
(Never commercialized)
Key players
Academia / research labs

Packet over SONET (PoS)

1996 – Backbone transport

How it workedMapped IP/PPP frames directly into a SONET payload for fast router-to-router links, dropping ATM's cell overhead.

Status
Dying
Killer / successor
10/40/100 Gigabit Ethernet
Key players
Cisco, Juniper

Semiconductor

Intel Optane / 3D XPoint

2017 – 2022 Persistent memory

How it workedA resistive, phase-change-like storage-class memory that sat between DRAM and flash - byte-addressable, persistent, and denser than DRAM.

Status
Dead
Killer / successor
Cost; CXL memory tiering
Key players
Intel, Micron

Itanium (IA-64 / EPIC)

2001 – 2021 CPU architecture

How it workedA VLIW-style 'EPIC' design that pushed instruction scheduling onto the compiler for wide parallelism - elegant in theory, brutal to compile for in practice.

Status
Dead
Killer / successor
x86-64 (AMD64)
Key players
Intel, HP

SPARC / MIPS / PA-RISC

1980 – 2021 CPU (proprietary RISC)

How it workedVendor-specific RISC instruction sets, each welded to a proprietary Unix and hardware line (Solaris/SPARC, HP-UX/PA-RISC, IRIX/MIPS).

Status
Dead
Killer / successor
x86-64; then Arm / RISC-V
Key players
Sun/Oracle, HP, MIPS/SGI

DEC Alpha

1992 – 2007 CPU (RISC)

How it workedA clean 64-bit RISC architecture that was, for much of the 1990s, the fastest processor on Earth.

Status
Dead
Killer / successor
Itanium strategy; x86
Key players
DEC, then Compaq/HP

EPROM (UV-erasable)

1971 – 2000 Memory

How it workedFloating-gate ROM erased by shining ultraviolet light through a quartz window, then reprogrammed on a special programmer.

Status
Dead
Killer / successor
Flash / EEPROM
Key players
Intel

RDRAM (Rambus DRAM)

1999 – 2000 Memory

How it workedA narrow, very high-speed serial memory channel that Intel pushed as the Pentium 4 standard - but it ran hot, cost more, and carried heavy licensing baggage.

Status
Dead
Killer / successor
DDR SDRAM
Key players
Rambus, Intel

ECL / bipolar mainframe logic

1960 – 1990 Logic

How it workedEmitter-coupled logic switched non-saturating bipolar transistors for raw speed, at the cost of running hot and power-hungry - used where speed beat all else (mainframes, Cray supercomputers).

Status
Dead
Killer / successor
CMOS
Key players
IBM, Cray, Motorola

GaAs digital logic

1980 – 1990 Logic / materials

How it workedGallium arsenide's higher electron mobility promised faster chips than silicon and powered the ill-fated Cray-3.

Status
Dead
Killer / successor
Silicon CMOS scaling
Key players
Cray-3, Vitesse

Transputer (INMOS)

1984 – 1990 CPU

How it workedA microprocessor with built-in serial links and on-chip memory, meant to be wired into large parallel arrays and programmed in the occam language.

Status
Dead
Killer / successor
Commodity microprocessors
Key players
INMOS (UK)

Josephson junction computing

1970 – 1983 Logic (superconducting)

How it workedSuperconducting junctions switched in picoseconds at liquid-helium temperatures, aiming at an ultra-fast computer.

Status
Dead
Killer / successor
(Cancelled) - CMOS progress
Key players
IBM

Magnetic bubble memory

1970 – 1980 Memory

How it workedStored bits as tiny magnetic domains ('bubbles') shuffled around a garnet film by a rotating magnetic field - solid-state and non-volatile, with no moving parts.

Status
Dead
Killer / successor
Cheap DRAM and disk
Key players
Bell Labs, Intel, TI

Magnetic core memory

1953 – 1975 Memory

How it workedTiny ferrite rings threaded on a wire grid stored each bit as a magnetization direction; reading a core flipped it, so the value had to be written back.

Status
Dead
Killer / successor
DRAM
Key players
IBM, MIT (Jay Forrester)

CCD image sensors

1969 – Sensor

How it workedCharge-coupled devices shuffled photo-generated charge packet-by-packet across the chip to a single low-noise readout amplifier.

Status
Dying
Killer / successor
CMOS image sensors
Key players
Bell Labs, Sony, Kodak

Enterprise

Novell NetWare

1983 – 2010 Network OS

How it workedThe dominant PC file-and-print network operating system, built on IPX/SPX with the fast NetWare Core Protocol.

Status
Dead
Killer / successor
Windows Server / IP
Key players
Novell

Dial-up Remote Access Servers (RAS)

1990 – 2000 Enterprise access

How it workedRacks of modems (or their digital equivalents) answered employee and ISP dial-in calls and bridged them onto the LAN or internet.

Status
Dead
Killer / successor
Broadband VPN
Key players
US Robotics, Ascend, Shiva, Cisco

H.320 ISDN videoconferencing

1990 – 2000 Enterprise video

How it workedRoom systems bonded several ISDN B-channels to scrape together enough bandwidth, coding video with H.261 over circuit-switched lines billed by the minute.

Status
Dead
Killer / successor
IP video (H.323/SIP), then cloud
Key players
PictureTel, Polycom, VTEL

IBM twinax / 3270 coax terminals

1970 – 2000 Terminal networking

How it workedDumb terminals hung off mainframes and AS/400s over coax (3270) or twinaxial cable (5250), daisy-chained back to cluster controllers.

Status
Dead
Killer / successor
Ethernet + PCs / emulation
Key players
IBM

ATM to the desktop / LANE

1995 – 1990 LAN

How it workedPushed 25/155 Mbit ATM cells all the way to the PC, with LAN Emulation faking a broadcast Ethernet over connection-oriented ATM so ordinary apps would run.

Status
Dead
Killer / successor
Switched Fast/Gigabit Ethernet
Key players
ATM Forum, Fore Systems

Bisync (BSC)

1967 – 1990 Data-link protocol

How it workedIBM's half-duplex, character-oriented Binary Synchronous protocol for batch and remote-job-entry links, later succeeded by bit-oriented SDLC.

Status
Dead
Killer / successor
SDLC/SNA, then IP
Key players
IBM

Centrex

1960 – Enterprise voice

How it workedPBX-like business features - extensions, transfer, hunt groups - delivered from the telco central office instead of on-site equipment.

Status
Dying
Killer / successor
IP-PBX / cloud voice
Key players
Bell operating companies

IBM SNA / SDLC

1974 – Mainframe networking

How it workedA hierarchical, connection-oriented mainframe network: SDLC on the wire, front-end processors and cluster controllers driving 3270 terminals, every path centrally predefined.

Status
Undead
Killer / successor
TCP/IP
Key players
IBM

TDM / digital PBX

1970 – Enterprise voice

How it workedOn-premises circuit-switched telephone exchanges wiring digital handsets over dedicated voice cabling, with TDM trunks out to the PSTN.

Status
Dying
Killer / successor
IP-PBX / cloud UC
Key players
Nortel (Meridian), Avaya (Definity), Siemens

X terminals / early thin clients

1990 – Desktop

How it workedDiskless graphical terminals ran the X11 display server and drew windows served from Unix hosts across the LAN.

Status
Dying
Killer / successor
Cheap PCs, then web / VDI
Key players
NCD, Sun, Wyse

Networking / IP / Security

Skype protocol (original P2P)

2003 – 2025 VoIP

How it workedA proprietary peer-to-peer overlay that used user 'supernodes' to punch through NAT and route calls; Microsoft later tore it out and re-hosted Skype in data centers.

Status
Dead
Killer / successor
Cloud VoIP; Teams
Key players
Skype, Microsoft

DES / 3DES / RC4

1977 – 2023 Cipher

How it workedDES's 56-bit key eventually fell to brute-force in days; RC4's statistical biases leaked plaintext; 3DES just ran DES three times to stretch its life.

Status
Dead
Killer / successor
AES
Key players
IBM/NSA, RSA

SSL 2/3 & TLS 1.0/1.1

1995 – 2021 Transport security

How it workedThe early web-encryption handshakes, later broken by attacks like POODLE and BEAST and dependent on obsolete ciphers (RC4, 3DES).

Status
Dead
Killer / successor
TLS 1.2 / 1.3
Key players
Netscape, IETF

IPv6 transition kludges (6to4/Teredo/NAT-PT)

2000 – 2015 Transition

How it workedTunneled IPv6 packets over the IPv4 internet (6to4, Teredo) or translated between the two protocols (NAT-PT) to bridge the transition.

Status
Dead
Killer / successor
Native dual-stack; NAT64
Key players
IETF

Minitel / Videotex

1982 – 2012 Online service

How it workedA free terminal in French homes dialed videotex services over the phone line for banking, directory, messaging and shopping - a national online service a decade before the web. Peaked near 9M terminals in 2002.

Status
Dead
Killer / successor
World Wide Web
Key players
France Telecom / PTT

PPTP VPN

1996 – 2012 VPN

How it workedTunneled PPP inside GRE, authenticating and encrypting with MS-CHAP and MPPE - both later shown to be badly breakable.

Status
Dead
Killer / successor
IPsec / OpenVPN / WireGuard
Key players
Microsoft

Cell Broadband Engine

2005 – 2010 CPU

How it workedOne PowerPC core plus eight SIMD 'SPE' accelerators on a die - powerful, but brutally hard to program (it powered the PS3).

Status
Dead
Killer / successor
GPUs
Key players
IBM, Sony, Toshiba

Cisco FabricPath

2010 – 2010 Data-center fabric

How it workedCisco's proprietary pre-standard take on the same idea: an IS-IS-routed Layer 2 fabric within the data center.

Status
Dead
Killer / successor
VXLAN / EVPN (ACI)
Key players
Cisco

Direct-detect 10G + dispersion compensation

1996 – 2010 Optical transmission

How it worked10G on-off-keyed light needed spools of dispersion-compensating fiber and careful per-span tuning to undo the chromatic dispersion that smeared pulses over distance.

Status
Dead
Killer / successor
Coherent DWDM (DSP)
Key players
Long-haul optical vendors

PATA / IDE

1986 – 2010 Storage interconnect

How it workedA 40- or 80-conductor parallel ribbon cable wiring two drives per channel, distinguished by master/slave jumpers.

Status
Dead
Killer / successor
SATA
Key players
Western Digital, ANSI

Shortest Path Bridging (802.1aq)

2012 – 2010 Data-center fabric

How it workedThe IEEE rival to TRILL, also IS-IS-based, carrying Ethernet inside Ethernet (MAC-in-MAC) across a loop-free multipath fabric.

Status
Dead
Killer / successor
VXLAN / EVPN
Key players
IEEE, Avaya/Extreme

TRILL

2011 – 2010 Data-center fabric

How it workedPut IS-IS routing underneath Layer 2 so a data-center fabric could use every link (no spanning-tree blocking), shortest-path-bridging MAC frames across the mesh.

Status
Dead
Killer / successor
VXLAN / EVPN
Key players
IETF, Cisco, Brocade

FB-DIMM

2006 – 2008 Memory

How it workedPut an Advanced Memory Buffer chip on each DIMM and serialized the memory channel to scale capacity - at the cost of extra latency, heat, and power.

Status
Dead
Killer / successor
Registered DDR3
Key players
Intel, JEDEC

WEP (Wired Equivalent Privacy)

1999 – 2004 Wi-Fi security

How it workedEncrypted Wi-Fi with the RC4 cipher and a tiny, reused initialization vector; the IV weakness let an attacker collect enough packets to recover the key in minutes.

Status
Dead
Killer / successor
WPA2 / WPA3
Key players
IEEE 802.11, Wi-Fi Alliance

Ethernet hub / shared CSMA-CD

1980 – 2000 LAN

How it workedEvery station shared one collision domain; a repeater hub blasted each incoming bit out every port, and CSMA/CD made stations listen, transmit, detect the collision, back off a random interval, and retry.

Status
Dead
Killer / successor
Full-duplex Ethernet switches
Key players
DEC/Intel/Xerox, Bob Metcalfe

Fibre Channel Arbitrated Loop (FC-AL)

1995 – 2000 Storage interconnect

How it workedWired up to 126 devices into a shared loop where they arbitrated for access - cheaper than a fabric, but shared bandwidth and one failure domain.

Status
Dead
Killer / successor
Switched FC fabric
Key players
INCITS T11

HiperLAN / HiperLAN2

1996 – 2000 Wireless LAN

How it workedEurope's WLAN standard; HiperLAN2 used 5 GHz TDMA/TDD with built-in QoS and a cellular-style MAC - technically strong for voice and video.

Status
Dead
Killer / successor
Wi-Fi (802.11a/g)
Key players
ETSI

HIPPI / GSN

1987 – 2000 HPC interconnect

How it workedA very wide parallel copper channel (HIPPI-800, then 6400/GSN) linking supercomputers to storage and peripherals at then-unmatched speed.

Status
Dead
Killer / successor
InfiniBand / Ethernet
Key players
ANSI, Cray, SGI

Parallel SCSI

1986 – 2000 Storage interconnect

How it workedA shared parallel bus with up to 15 devices, terminators at each end, and ever-tighter cable-length limits as clock rates climbed.

Status
Dead
Killer / successor
SAS (Serial Attached SCSI)
Key players
INCITS T10

X.400 (Message Handling)

1984 – 2000 Email

How it workedThe OSI email standard, with rigid hierarchical addresses (C=, O=, OU=, S=) and carrier-run store-and-forward message transfer agents.

Status
Dead
Killer / successor
SMTP / Internet email
Key players
ITU-T/ISO, PTTs

X.500 / DAP

1988 – 2000 Directory

How it workedA global hierarchical directory reached through the heavy Directory Access Protocol running over a full OSI stack.

Status
Dead
Killer / successor
LDAP
Key players
ITU-T/ISO

BITNET

1981 – 1996 Academic net

How it workedAn academic store-and-forward network linking mainframes over leased lines - home of LISTSERV and early cross-campus file and message exchange.

Status
Dead
Killer / successor
The Internet
Key players
Universities (IBM RSCS/NJE)

Classful addressing / RIPv1

1981 – 1993 Routing

How it workedIP addresses came in fixed Class A/B/C sizes, and RIPv1 advertised routes with no subnet mask - wasting address space and bloating routing tables.

Status
Dead
Killer / successor
CIDR; RIPv2 / OSPF
Key players
IETF

Archie / Veronica

1990 – 1990 Internet search

How it workedArchie indexed the file listings of anonymous-FTP servers; Veronica did the same for Gopher menus - the first tools to search the internet.

Status
Dead
Killer / successor
Web search engines
Key players
McGill University

EGP (Exterior Gateway Protocol)

1982 – 1990 Routing

How it workedThe original inter-autonomous-system reachability protocol, which assumed the early internet was a simple tree with a single backbone core.

Status
Dead
Killer / successor
BGP
Key players
IETF

FidoNet

1984 – 1990 BBS network

How it workedA global network of dial-up bulletin-board systems that relayed echomail and netmail overnight on a scheduled store-and-forward 'mail hour.'

Status
Dying
Killer / successor
The Internet
Key players
BBS hobbyists (Tom Jennings)

Prestel / Viewtron / Teletext

1979 – 1990 Info service

How it workedVideotex (interactive, over phone lines) and teletext (one-way, broadcast in the TV signal's vertical-blanking interval) painted pages of text and blocky graphics onto TV screens.

Status
Dead
Killer / successor
The Web
Key players
British Telecom, Knight-Ridder/AT&T

SSA (Serial Storage Architecture)

1991 – 1990 Storage interconnect

How it workedIBM's dual-path serial storage loop - technically ahead of early Fibre Channel, but proprietary.

Status
Dead
Killer / successor
Fibre Channel
Key players
IBM

UUCP

1979 – 1990 Store-and-forward net

How it workedUnix-to-Unix Copy dialed machine-to-machine, usually at night, to relay email and Usenet in store-and-forward hops addressed by bang paths (host!host!user).

Status
Dead
Killer / successor
TCP/IP Internet
Key players
AT&T / Bell Labs

Intel iAPX 432

1981 – 1986 CPU

How it workedAn ambitious object-oriented, capability-based processor so complex it ran slower than the cheap 8086 Intel shipped as a stopgap.

Status
Dead
Killer / successor
x86 (the 8086 stopgap)
Key players
Intel

Group 3 fax (T.30) / T.38

1980 – Fax

How it workedEncoded a scanned page as audio tones over a voice call; T.38 later relayed those same fax tones across IP networks so fax could survive VoIP.

Status
Dying
Killer / successor
Email / scan-to-PDF; secure messaging
Key players
ITU-T

NetBIOS / WINS name resolution

1987 – Naming

How it workedFlat, broadcast-based name resolution for Windows networks; WINS added a central server to map NetBIOS names to IP addresses without broadcasts.

Status
Dying
Killer / successor
DNS
Key players
IBM, Microsoft

Space

LORAN-C

1957 – 2010 Radionavigation

How it workedChains of low-frequency ground transmitters sent timed pulses; receivers fixed position from the time differences between master and secondary stations (hyperbolic navigation).

Status
Dead
Killer / successor
GPS
Key players
US Coast Guard

Inmarsat-A

1982 – 2007 Satellite telephony

How it workedThe first satellite phone service: analog FDM voice relayed via geostationary satellites to ship-borne one-meter gyro-stabilized dish antennas.

Status
Dead
Killer / successor
Digital Inmarsat-B/Fleet; VSAT
Key players
Inmarsat

SkyBridge

1997 – 2002 Satellite broadband

How it workedA planned 80-satellite Ku-band LEO broadband constellation that never got off the ground.

Status
Dead
Killer / successor
(None) - never launched
Key players
Alcatel

PrimeStar

1994 – 2001 Satellite TV (DBS)

How it workedMedium-power Ku-band direct broadcast needing a ~1 m dish, offered on a rent-the-equipment model.

Status
Dead
Killer / successor
High-power DBS (DirecTV)
Key players
Cable-company consortium, GE

Analog C-band TVRO

1979 – 2000 Satellite broadcast

How it workedTwo-to-three-meter backyard dishes pulled analog FM television feeds straight off C-band satellites (later scrambled with VideoCipher).

Status
Dead
Killer / successor
Digital DBS (DirecTV/Dish)
Key players
HBO, backyard-dish owners

Celestri

1997 – 1998 Satellite broadband

How it workedA proposed 63-satellite LEO-plus-GEO broadband constellation; folded into Teledesic before any launch.

Status
Dead
Killer / successor
(None) - folded before launch
Key players
Motorola

Omega

1971 – 1997 Radionavigation

How it workedEight very-low-frequency stations gave worldwide position fixes by comparing signal phase between stations.

Status
Dead
Killer / successor
GPS
Key players
US Navy + partners

Transit

1964 – 1996 Satellite navigation

How it workedThe first satellite navigation system: low-orbit satellites broadcast a stable tone, and a receiver computed its position from the Doppler shift as the satellite passed overhead - roughly one fix every 90 minutes.

Status
Dead
Killer / successor
GPS
Key players
US Navy, Johns Hopkins APL

Moonshots & Vanished

Facebook Express Wi-Fi

2016 – 2022 Public Wi-Fi

How it workedPartnered with local ISPs and small entrepreneurs to sell affordable Wi-Fi hotspots in emerging markets.

Status
Dead
Killer / successor
Strategy shift
Key players
Meta

Meta Connectivity / Terragraph

2016 – 2022 60 GHz fixed wireless

How it workedTerragraph used 60 GHz (802.11ay) rooftop nodes for cheap urban fixed wireless; Meta Connectivity was the umbrella over Aquila, Express Wi-Fi, Magma and more.

Status
Dead
Killer / successor
Refocus; licensed out
Key players
Meta

Google/Alphabet Project Loon

2011 – 2021 Stratospheric balloons

How it workedSolar-powered balloons drifting in the stratosphere (~20 km), steering by rising and falling into different wind layers, relaying LTE from ground stations to phones far below.

Status
Dead
Killer / successor
Unit economics
Key players
Alphabet X

Google Station

2016 – 2020 Public Wi-Fi

How it workedFree, ad-supported public Wi-Fi at Indian railway stations and other venues across several countries.

Status
Dead
Killer / successor
Cheap mobile data (Jio)
Key players
Google

OneWeb (v1)

2012 – 2020 LEO broadband

How it workedA planned ~650+ satellite low-earth-orbit constellation for global broadband, a direct Starlink rival.

Status
Resurrected
Killer / successor
(Rescued) - SoftBank pulled funding
Key players
Greg Wyler, SoftBank; then UK govt/Bharti/Eutelsat

LeoSat

2013 – 2019 LEO enterprise data

How it workedA proposed laser-linked LEO constellation for secure enterprise and backhaul data - not consumer broadband.

Status
Dead
Killer / successor
Lost investors
Key players
LeoSat Enterprises

Facebook/Meta Aquila

2014 – 2018 Solar drone

How it workedA 737-wingspan, feather-light solar drone designed to circle at altitude beaming connectivity down for months at a time.

Status
Dead
Killer / successor
Partner-led HAPS strategy
Key players
Facebook Connectivity

Google Titan (solar drones)

2014 – 2017 HAPS drone

How it workedHigh-altitude, solar-powered drones meant to loiter aloft for months as atmospheric cell towers.

Status
Dead
Killer / successor
Refocused on Loon
Key players
Google / Alphabet

Facebook Free Basics / Internet.org

2013 – 2016 Zero-rated access

How it workedOffered a free, curated slice of the internet - Facebook plus selected partner services - on partner carriers across developing markets.

Status
Dead
Killer / successor
Net-neutrality backlash
Key players
Facebook

Google Fiber

2010 – 2016 Fiber-to-the-home

How it workedGigabit fiber-to-the-home in select cities, launched to pressure incumbent ISPs on speed and price.

Status
Undead
Killer / successor
Build economics
Key players
Google / Alphabet

LightSquared

2010 – 2015 LTE / satellite hybrid

How it workedPlanned a nationwide LTE network reusing L-band satellite spectrum on the ground; the FCC blocked it when tests showed it swamped adjacent GPS receivers.

Status
Dead
Killer / successor
GPS-interference ruling
Key players
Philip Falcone / Harbinger

Aereo

2012 – 2014 Streaming broadcast TV

How it workedRented each subscriber a tiny individual antenna in a data center to capture and stream local over-the-air TV over the internet.

Status
Dead
Killer / successor
Supreme Court copyright ruling
Key players
Aereo (Chet Kanojia)

WebTV / MSN TV

1996 – 2013 Internet-on-TV

How it workedA set-top box that put a dial-up web browser and email onto an ordinary television set.

Status
Dead
Killer / successor
Smartphones / smart TVs
Key players
WebTV Networks, Microsoft

Municipal Wi-Fi (EarthLink/Philadelphia)

2005 – 2008 Public Wi-Fi

How it workedCities and EarthLink blanketed downtowns with mesh Wi-Fi nodes on light poles to offer cheap or free public internet.

Status
Dead
Killer / successor
Economics / 3G
Key players
EarthLink, US cities

AT&T Project Angel

1997 – 2002 Fixed wireless local loop

How it workedDelivered home phone and broadband over fixed wireless - a rooftop antenna to a neighborhood base station - to bypass the incumbent's copper local loop entirely.

Status
Dead
Killer / successor
DSL / cable; later WiMAX
Key players
AT&T Wireless (McCaw), Motorola

Global Crossing

1997 – 2002 Undersea fiber

How it workedBuilt a worldwide undersea-and-terrestrial fiber network on the conviction that bandwidth demand was effectively limitless.

Status
Dead
Killer / successor
Bandwidth glut / crash
Key players
Gary Winnick

DSL CLECs (NorthPoint, Rhythms, Covad)

1997 – 2001 DSL access

How it workedColocated their own DSLAMs inside incumbent central offices to sell wholesale and retail DSL under regulatory line-sharing rules.

Status
Dead
Killer / successor
ILEC competition / crash
Key players
NorthPoint, Rhythms, Covad

Enron Broadband Services

1999 – 2001 Bandwidth trading

How it workedAimed to trade network bandwidth as a commodity and stream video-on-demand over an 'intelligent' content network.

Status
Dead
Killer / successor
Fraud / collapse
Key players
Enron

Excite@Home (@Home Network)

1996 – 2001 Cable broadband

How it workedPioneered national high-speed cable-modem internet as a service layered over cable operators' plants, bundled with the Excite web portal.

Status
Dead
Killer / successor
Cable MSOs (insourced broadband)
Key players
TCI, Comcast, Cox, Excite

Teligent / Winstar

1997 – 2001 Fixed wireless CLEC

How it workedUsed millimeter-wave rooftop radios (24/28/38 GHz) to deliver competitive business broadband, bypassing the incumbent's wires.

Status
Dead
Killer / successor
Dot-com crash / economics
Key players
Teligent, Winstar

PointCast Network

1996 – 1999 Push media

How it workedA screensaver that 'pushed' news and stock tickers to desktops over the corporate LAN - the poster child of 1990s push technology.

Status
Dead
Killer / successor
Bandwidth strain; RSS/notifications
Key players
PointCast

Death dates for standards are editorial judgments, not ratified events. Corrections and remembrances are welcome — if you helped build or bury one of these, we would like to hear from you.

Visit the Company Graveyard → Explore 190 years of milestones →