What are the real-world consequences of these attacks for users and owners of wireless networks
Already a great answer here, but thought I would add my viewpoint to a part of it. There have been a lot of "sensationalist" headlines and misinformation in recent days that portray this vulnerability as much more serious than it really is.
Ultimately, this vulnerability, while very serious, will have very little impact on the day to day of most users and I don't expect to see this exploit much "in the wild." Frankly, there are far too many open networks that are much easier to exploit for an attacker to gather personal information.
The attack vector using KRACK is simply too small (and will continue to decrease) to make these attacks widespread. There are 10 CVEs associated to this vulnerability, 9 client related and 1 infrastructure related. Patching the infrastructure mitigates 8 of the CVEs (including the most serious) mainly leaving client-to-client connections vulnerable (when was the last time you used and ad-hoc or Wi-Fi Direct 802.11 connection?). Patching the client mitigates all but the infrastructure CVE. Don't want to patch the OS? Patching even the network driver on the client will mitigate at least 2 of the CVEs.
Two of the biggest target operating systems, Windows (7+) and iOS (10.3.1+), were not vulnerable on day 0 unless on a network with 802.11r (fast roaming/transition) enabled. Of those two, Windows already had a patch released over a week ago. Patches are also out for most of the common flavors of Linux and in beta for all Apple OSes. You can expect most of the current mainstream operating systems (and nearly all the Linux variants) to have a patch released within a couple weeks. All in an age where OS upgrades are easier and more automated than ever.
This leaves the legacy operating systems and IoT to consider. Fifteen to twenty years ago, legacy devices would have been much more of a concern but today with commodity electronics that are much cheaper and often replaced every couple years (if they last that long), we have a much lower percentage of "old" devices hanging around. For the IoT? If you really want to watch my lights (or whatever) turn off and on, please feel free. Yes there is potential for more meat on the IoT bone, but mainly only in very limited corner cases and not to the average user.
When it comes to 802.11r, most consumer access points (aka "routers" by many) simply do not support 802.11r. Vendors tend to see little value in adding support for it when the majority of their equipment is deployed to environments where it is the only wireless AP. Single AP means no roaming which certainly precludes the need for fast roaming and also means no patch needed. Of the ones that I have seen that support it, most have 802.11r disabled by default (primarily due to issues some clients who don't support 802.11r have with it).
802.11r is much more widespread in multi-AP deployments and most of the common vendors for such environments (Cisco, Aruba, Ubiquiti, Ruckus, Aerohive, etc) have patches out already for some or all of their devices. These are also the environments that are more likely to have paid staff or support consultants that are aware of this exploit.
Many "high value" targets are also out as they enforce the use of multiple layers of encryption when using wireless. Yes, you can break the 802.11 encryption, but not the VPN encryption in use on the connection or the HTTPS traffic within the VPN tunnel. Targets that depend on keeping their data secure aren't trusting to encryption that only covers from the client to the AP.
Even targets that aren't high value are often using other encryption without any change in behavior. Most "big" websites already push all their traffic to HTTPS as do most sites handling any sort of financial or personal information.
To perform many types of MitM attacks (which really require bidirectional control), an attacker needs to have targets that are either using GCMP or have both using 802.11r and clients with the 4-way handshake vulnerability. GCMP isn't common yet and we have already hit on 802.11r and client patching. So while the MitM demonstration shown as a proof of concept is impressive, the real world implications are fairly limited.
If you understand this vulnerability enough to exploit it successfully, you will quickly realize what I already mentioned above....it is much easier to exploit the many open wireless networks that exist all around us.