I should have known any company with a parent that's a telco could not be trusted. Just to recap, I've been trying since October to get a minimally tolerable level of service from them, or at least some kind of acknowledgment that they have a problem and some effort to take responsibility for their services. Once I finally did get someone who was interested in helping, so they laid him off. Their support people are incompetent, and blame everything on nothing, searching for any excuse to dismiss my problem as either imagined, nonexistent, impossible, or some combination of the three. After months of dealing with this, I've decided that future attempts to contact their personnel either by phone or email is going to be utterly useless. Their support lackeys have no interest in fixing whatever it is that's gone wrong, especially since they sternly refuse to admit that something is wrong. It's not possible to bypass their lower primate customer service personnel to talk to someone who actually has a clue (if, in fact, Sprint employs any such people). My last resort is to contact someone at Sprint corporate and try to get my request for real assistance in from the top rather than the bottom.

To that end, I make the following accusation, and leave it to Sprint to prove me wrong. Sprint is actively defrauding its customers. What exactly is fraud, you ask? I found this definition online:

Fraud is defined to be "an intentional perversion of truth" or a "false misrepresentation of a matter of fact" which induces another person to "part with some valuable thing belonging to him or to surrender a legal right".

So now, I will illustrate for you how Sprint has perverted the truth and misrepresented facts, causing me to surrender my valuable time and money.

The best way to do this is with Sprint Fraudband Defect's own words. Here are things they've said about their service, coupled with my own experiences.

With its "always-on" connection, customers commonly experience download speeds of 512 thousand bits per second (kbps) to 1.5 million bits per second (1.5 Mbps), with burst rates up to 5 Mbps. With such lightning-fast speeds, business and residential customers can take full advantage of Internet offerings that require a high-speed connection, such as streaming video and graphic-intensive Web sites.

---Sprint Broadband press release

My own tests have shown this not to be the case. See the articles linked above for some examples. At the moment, my downstream speed is about 420kbps and has been between 310kbps and 440kbps for the last hour. How do I know this? I have a perl script testing the speed for me at regular intervals. What Sprint doesn't tell you is that for the first 320kB of a file, you get lousy speeds. Their equipment allows customers to "monopolize" a channel (called "dedicated state") that's being heavily used, so after 320kB, you see a dramatic improvement in speed. Knowing this, they designed their speed testing web site to send a couple megs of data, which skews the apparent speed. As most people are aware, not everything you download is more than 320kB. Most web pages are way smaller than that. Most morons.org web pages are under 20kB, for example. Sprint makes NO mention of this caveat on their web site.

Sprint Broadband Direct beams the Internet directly into homes and businesses at download speeds up to 50 times faster than a standard dial-up modem....With its "always-on" connection, customers commonly experience download speeds of 512 thousand bits per second (kbps) to 1.5 million bits per second (1.5 Mbps), with burst rates up to 5 Mbps. With such lightning-fast speeds, business and residential customers can take full advantage of Internet offerings that require a high-speed connection, such as streaming video and graphic-intensive Web sites.

---Sprint Broadband press release

Maybe it's 50 times faster than a teletype. Under peak usage conditions, Sprint Broadband's connectivity is ONLY useful for those streaming video and graphic-intensive web sites, for the reasons I stated above.

One of the things they tell me time and time again when I report problems is that they don't support any minimum latency. Latency is simply the round-trip-time to send one packet of information. For a dialup modem, sending a packet to the first router and back averages around 100ms these days. During peak periods of usage, the round trip time to my Sprint Broadband gateway is often over 500ms. Mind you, we're just sending a 56-byte packet there and back. So if it takes 500ms to transmit 56 bytes, that's a lousy 28Bps, or roughly the speed of a 300 baud modem. Mind you, it's rare that we'd only be sending 56 bytes somewhere, but is it really reasonable for it to take this long to send one small packet across a supposedly "fast" connection? Presumably since they claim not to support any minimum latency, it would be OK by Sprint if my packet took an hour, a day, or a week to return. Most people would call that unreasonable, but Sprint shrugs it off. Clearly for small payloads, Sprint Fraudband is not 50 times faster, but 5 times slower than a dialup modem. In fact, it isn't until a download becomes at least 64k or so that the high latency stops degrading the speed of transmission by such a large ratio (that is, the connection set-up time ceases to be such a large percentage of the total transfer time. Once a TCP connection is set up, TCP uses a sliding-window algorithm to keep the "pipe" "full". That is, the sender will keep sending data without waiting for acknowledgment, making more efficient use of the bandwidth. Setting up the connection requires back-and-forth communication, which is highly latency-sensitive.).

I question their insistent unwillingness to acknowledge that supporting a minimum latency is essential. Their excuse is that "it's the nature of the technology". The fact that their service did not, prior to August, suffer from the type of degradation I've seen since August is evidence to the contrary. I have also consulted with an engineer at Terranova.net, a wireless ISP in the Florida keys, who told me that high latency is only a factor when access units are heavily oversubscribed; that is, access units are only able to manage a certain number of packets-per-second (pps) and if the access unit is getting hammered, packets will be queued and latency will be high. This information is, of course, ignored and dismissed by Sprint personnel with a wave of an excuse and a policy of the ostrich algorithm.

The fact that the usage patterns of customers is nondeterministic, combined with overloaded access units (or other problems) contributes to another networking phenomena called "jitter". Simply defined, jitter is variance in the latency on a connection. Most connections have fairly constant latency. That is, a dialup modem will pretty consistently have a round-trip time of 100ms or so. What happens when you have large variations in latency is packets get out of order upon arrival. Consider the following example:

Packet 1 leaves at time 0ms and encounters 10ms of latency.
Packet 2 leaves at time 10ms and encounters 10ms of latency.
Packet 3 leaves at time 20ms and encounters 100ms of latency.
Packet 4 leaves at time 30ms and encounters 10ms of latency.

The packets will then arrive at their destination at the following times:

Packet 1 arrives at time 10ms.
Packet 2 arrives at time 20ms.
Packet 4 arrives at time 40ms.
Packet 3 arrives at time 120ms.

Note that because of the jitter, the latency for the third packet was much higher, causing it to arrive out of order. Now imagine what happens if these packets are part of a video stream, or an audio stream for that matter. Either the buffer size has to be huge to ensure that the packets can all be re-ordered before they're played, or the out-of-order packet has to drop, degrading quality, or making the user wait a long time before the stream begins playing while the buffer fills.

Sprint owns its entire broadband network and manages its traffic for optimum performance 24 hours a day, 7 days a week.

---Sprint Broadband press release

I guess 24 hours a day somehow doesn't include the hours between 7 PM and 11 PM. See the articles linked in way at the top of this one.

* No more long wait times for large graphics and web pages to download

---Sprint Broadband press release

We've all heard the term "World Wide Wait" by now. Part of what causes the "world wide wait" is the TCP handshake process at the opening of a web connection. As I've already noted here, the TCP handshake is highly latency-sensitive. Out of curiosity tonight I've been timing how long it takes to request a 0-byte file from the morons.org web server. This sort of request would essentially only include the "wait" part of the "world wide wait" since there's no page data to download. From my Sprint Fraudband connection, one HTTP request is taking anywhere from .36 seconds to well over a second. Compare that to .1 seconds from the web server itself, and .13 seconds on average from one of my shell accounts on a T1. So the very minimum time it could take to download a page over my Sprint Fraudband connection is .36 seconds, and that's for a page with absolutely no content. Considering that most web pages do have content, even loading up a simple text web page (like morons.org for example) could take several seconds. In fact, downloading the morons.org homepage takes me 4.5 seconds, and it's only 22kB. This is roughly the same amount of time it would take to download the page over a 56k modem. So much for "no more long wait time".

Sprint Broadband Direct is an innovative, new method of delivering high speed broadband Internet access to residential and business customers. It utilizes a new technology called Multi-channel Multi-point Distribution System (MMDS) that delivers broadband over a secure, always on, reliable fixed-wireless network at speeds up to 50 times faster than a 56K modem.

--- Sprint Broadband press FAQ

* Highly secure/Highly reliable.

--- Sprint Broadband "about" web page

Sprint Broadband Direct is a broadband access service. Broadband is a transmission technology that passes data with a bandwidth, or capacity, greater than your typical phone line.

The result: speedy, consistent service.

--- Sprint Broadband "technology" web page

Main Entry: reliable
Function: adjective
1 : suitable or fit to be relied on : DEPENDABLE
2 : giving the same result on successive trials

--- Merriam-Webster's online dictionary

Can you really call a network with the characteristics I've described here "highly reliable"? I've already demonstrated why their speed angle is a farce, at least for files under 320kB in size, and especially for files under 64kB in size. If it takes 10 seconds to get a character echoed back to me in an ssh session, would I call that "reliable", especially if 20 minutes before it might have been echoed back instantly? Is this an example of giving the same result after multiple trials then? I asked this of a support person once, and they claimed to me that their connectivity isn't meant for telnet and online gaming. Funny, there's no mention of that anywhere on their web site.

Would a reasonable person think that having to buffer up 768k of streaming mp3 data just to ensure playback without skipping the result of having a "reliable" connection? How about a "highly reliable" connection?

Would a "highly reliable" connection that's "monitored 24 hours a day" show significant degradation between certain hours of the day? What was that about giving the same results upon multiple trials? Maybe Sprint is working from some other definition of "reliable" that means "highly variable and undependable". I guess they have different definitions for "speedy" and "consistent" as well.

I think it's pretty clear that we have an "an intentional perversion of truth" here. I've gone to great pains to show Sprint that they do indeed have problems. Sprint knows about this. Sprint chooses to do nothing about it, all the while making the same claims that their service is fast, reliable, etc.

I also think it's clear that there is "false misrepresentation of a matter of fact" here. Sprint Fraudband advertises that they have fast speeds, but doesn't say you'll only get those speeds if you're downloading at least 320kB of data. They make no mention of the fact that their latency and jitter will kill ssh, telnet, online gaming, chatting, and similar applications. They go so far as to claim their service is great for streaming video, when we can easily see that huge amounts of jitter severely handicap realtime applications of this nature. Sprint has lied about their service, and when they haven't lied about it, they've only told half of the truth.

Fortunately for consumers, Sprint Fraudband has stopped taking orders for their service, so no new people will be harmed by their dishonest practices. Existing customers who are locked into 1-or-2-year contracts, however, will continue to suffer service that has degraded well below the lofty promises Sprint made to them before they "parted with some valuable thing"... their time and money. Most Sprint Fraudband customers paid $99 or $199 for their hardware (depending on the length of their contracts) and are now paying $40-50 a month for their service. They've clearly parted with their valuable time. It's impossible to calculate how much time customers like myself have wasted on the phone with incompetent technical support or exchange email with those primitive drones. Time is also a valuable thing.

It is therefore my position that Sprint Fraudband Defect is actively engaged in willful fraud, to say nothing of despicable customer care. I'm keeping my eyes and ears open to see if someone starts a class action against this deceitful company. What I really want, however, more than anything else, is for Sprint Fraudband Defect to listen, acknowledge the problem, accept responsibility, and most importantly, fix it.

---Nick