"There can be at most one successful protocol for a given use case."
-
@evan yeah, they're both supported in hundreds of millions of devices worldwide. I've seen furnace filters with BLE. There's a huge market share for Zigbee in IoT/smart home and industrial applications. You'll find 2.4GHz propRF in electronic shelf labels, utility meters, and sensors among other use cases.
-
BeAware :fediverse:replied to Evan Prodromou last edited by
@evan this one's nuanced for me.
I am somewhat disagree but only if the protocols play nice.
In the social media space, it seems as though ActivityPub and protocols that proceeded it that you were also a part of, came first. (Correct me if I'm wrong, I barely know about all the different protocols)
If that's the case, it's really on the other protocols to play nice and show that they support open communication.
However, that's not happening, so I lean towards agree.
-
@evan I slept in it and came up with another couple of examples. For more than two decades we had CDMA and GSM as differing protocols for cell phones in the US. For satellite positioning there's GPS and GLONASS (depending on how you want to define protocol since those are both one way signals)
-
@evan
(cough)
Programming languages… -
@Affekt so, I'd give IOT as an example of a problem area where too many incompatible protocols has inhibited growth of the market.
-
@Affekt did it help to have multiple protocols for phones in the US?
-
@badibulgator are programming languages protocols?
-
@evan We could discuss definitions but… same kind of thing surely, no?
-
@luis_in_brief when humans had one language, we built a tower to Heaven, and God had to strike it down and confuse our tongues or we'd get too powerful.
-
Mx. Aria Stewartreplied to Evan Prodromou last edited by
@evan Making me think real strongly about network effects — I want to say disagree, but there _are_ major forces making that not entirely be the case.
-
Evan Prodromoureplied to Mx. Aria Stewart last edited by
@aredridel I tell anyone I can about network effects. If the value of a network is in the possible connections (Metcalfe's Law), the value of a network with N nodes varies with N^2. That's why big networks succeed.
If you take a total population of nodes N and split it into two incompatible networks of equal size N/2, the value of each is proportional to (N/2)^2 or N^2/4. If you add up the value, it's N^2/2 -- about half of what the value would be with one united network.
-
Ben Thompson 🐕replied to Evan Prodromou last edited by
@evan knives do not necessarily have to go on the right side of the plate. Lefthanders benefit from their being two protocols.
-
Evan Prodromoureplied to Ben Thompson 🐕 last edited by
@jbenjamint I'd say the protocol there is "put the knife on the side of the dominant hand"
-
@evan I am going to strongly agree with this. Having to understand multiple protocols for the same use case is a lot of mental load and makes the abstraction of the use case leakier. Multiple protocols also requires a bunch more development decisions: deciding which protocol is "best", which library is most robust, etc. This all takes away from the effort of designing the larger scale system of which the use case is a smaller part of.
-
Mx. Aria Stewartreplied to Evan Prodromou last edited by
@evan Exactly. And yet, there's problems of scale, so the equation isn't quite so simple, and trying to make one network do everything is perhaps even more limiting.
-
Evan Prodromoureplied to Mx. Aria Stewart last edited by
@aredridel and yet we have the Web and email.
-
Chris (Master of Potate) 🥔replied to Evan Prodromou last edited by
@evan HDMI vs. DisplayPort. DirectX vs. Vulcan. Matrix vs. Signal. It also depends on how broad you define use case and protocol.
-
Evan Prodromoureplied to Chris (Master of Potate) 🥔 last edited by
@chris those are 3 examples of one successful thing and one barely surviving thing!
-
@badibulgator I'm in a program called Summer of Protocols right now and we talk about protocols in the abstract a lot so everything looks like a protocol.
-
@evan SMTP is great for sending mail but HTTP is perhaps a more modern approach. Likewise FTP. It surely helps if there’s a winning protocol but that can result in being stuck in a local maximum due to upgrade/switching costs. On the other hand, proliferation brings fragmentation. So option ‘E’ - “it’s complicated”