Reflections on the ~2005~ CQWWW Contest

Well, it's over now. I can look back over the last 48 hours and put a few thoughts down while they're fresh in my mind. It appears I've just won the 2005 CQWWW. I felt pretty good all weekend and knew I had a good shot at it. The screen was filling up smoothly all weekend and there didn't seem to be too many times when I wasn't making contacts, er, connections. It reminds me of the old days, especially when we had high sunspot counts to help the signals go from antenna to antenna. Now, of course, all that has changed.

Back at the turn of the millennium, when all electromagnetic emissive modes were declared obsolete for purposes other than radar and satellite links by International Communications Minister Nicholas Negroponte, contesting officially turned completely to the wired medium. This completed the Negropontization of society. For those of you who may have forgotten, when Negroponte was at the MIT Media Lab back in the 1980s, he predicted that a "flip" was coming, where all communications then carried on wires would go wireless, and all wireless communications would become wired. He cited TV, which had made the wireless-to- wired flip, and telephones, which had gone from wired to wireless. Of all the communications services known, amateur radio had been the last holdout to go completely wired. Nearly all the hams were off the air and on the wire by the time the government mandated the end of wireless, except for a few fossilized holdouts on 80M CW and 14313. These last few were hunted down and are now in prison, I think.

Anyway, since the late 90s, most of the interesting contests have taken place on the wire. From 4-hour trial events on the mid- 90s Internet to full blown 48-hour marathons like the CQWWW on the now fully hampopulated WorldWide Web (which CQ Magazine purchased several years ago...), there is now a contest almost every weekend. The casual operators complain about the slowdown in Web traffic when there's a contest on, but they can always go to another service like AOL or go watch Applevision on cable. Nobody says they have to get on the Web that weekend.

The first few wired contests were chaotic affairs, with some people new to the mode calling out of turn continuously and jamming the streams. I remember one contest, where some guy (I think he was in southern Europia) screwed up his protocol stack with careless editing and locked up one stream on the NSF backbone for the whole 48 hours transmitting and never receiving. Even though the transmission speed limit in his country was 300 kb/s, he had to be running at least 10Mb/s to tie up the whole link like that.

As software and operating skills improved, a whole lot of tricks emerged. Some teams in what used to be Oregon and Slovenia conducted experiments in completely- automated operation, but when these multi- site internetworked systems took over the whole net, they were outlawed. Now human presence is required on at least one keyboard of a station.

My own station is pretty modest, but I know how to get the most out of it. I have three machines here, two for connections, and one for spotting. The two connection machines are IBMwoods, and the spotting machine is my the spotting machine is my trusty old ICOMBell. I got it at Dayton a couple of years ago, and it's slow but does the job. I've put in all the mods from the NWCJ (National Wired Contest Journal) articles and a few of my own.

Outside, I have a bit of a problem. My location puts me at a disadvantage. Out here in sector NE01 (which used to be called New Hampshire), we seem to be on the trailing edge of connection technology installation. We just got fiber-to-the-home two years ago, and the links to the nearest major backbone are two switches away. Guys a bit closer to the coast and further south have a big advantage - probably 20 microseconds of transfer latency. That can mean the difference between connecting to a new multiplier immediately or waiting in queue for an hour while he clears the stack. The West Coast guys in Silicon Valley have a bigger advantage, with their access to all the latest efficient network equipment. Rumor has it that in the last contest, some of them were using some new commercial meta- neural-commware from giant TGV Corporation. It isn't strictly against the rules, and we hams have always been encouraged to experiment, but blatant use of expensive commercial software like that seems to violate the spirit of the rules to me. I guess I'm just too much of a traditionalist.

Anyway, I've learned to compensate for my station's shortcomings here in the Black Hole of communications by operating tricks. Now that we don't have to worry about ionospheric propagation, the trick is finding optimum network propagation. Like the old days, you have to also make sure you hit the population centers at times when most of the machines are likely to be on, usually daylight and early evening hours. There's always a lull in activity when the environmental danger report is broadcast (every morning at 0700 local), so people know what to wear for the day (whether people will need their ultraviolet-protection shields, for example).

Here's how it went this weekend. When the contest started, the ARRL and CQ subnets of the Web backbone were jammed as usual. Even though people were trying to run at 10Mb/s, all the collisions held everyone's rate down. I worked around the big mess, going instead to a small switch I found when cruising the Web a few weeks ago. Even though it's just a 1 Mb/s node, it's virtually vacant, and has access to a satellite shot out to the Far East. I got in, looked around, and there were no other contesters connected, just a couple of old timers talking about the weather on the switch - a local QSO. They were going pretty slow, like most bit-chewers, and I could multiplex at 90% share right into them. They didn't even notice me passing through. I had the node essentially to myself. I tickled the satellite port and woke it up. I set myself as the primary user, and started looking for downlink receptors. I found the node I wanted in sector AS-25A (a small suburb outside what used to be called Tokyo). I synced up quickly and set to work. First, a broadcast to all local logins announcing my presence and availability for connects. I then quickly downloaded my own home-brew compression protocol to the common area, and routed all return connections to me through the compressor. The node sprang to life just as the enviro report was ending. Four thousand users got my solicitation and decided to give me a quick connection. The returns starting going to disk immediately, with the decompressor running at my end in spare cycles. I chuckled to myself as I noticed the first few decompressions - all the files began with the traditional "59925" sync header. Just like the old days.

The connection hang in pretty well for about two hours, then I guess the sun got higher in the sky in '25A and users went about their recreations for the day. The rate dropped and I knew it was time to change to a different node. Before leaving, I planted a bug in the node that identified all contest file exchanges and inflated their sizes. I call it "defensive contesting". It's not against the rules yet, so it's OK for now, but there's talk in InterCongress about having the practice banned. Then I guess I'll have to find another trick. But it's getting harder to be creative since the logs are so closely checked now. With the return of prize money, I guess it was inevitable that checking would get more thorough.

Anyway, back to the contest. The ICOMBell checked the spotting nets and the WWV (World Weather Vote) numbers. Looked like the environment was going to be in pretty good shape in the major population areas, so people would be going outside a lot. Most metros were planning cool temperatures, by popular enviro-vote, so few people would be lounging around the pool with their portables handing out connections. Darn. This was shaping up to be a night-time weekend. I prefer the weekends with round- the-clock activity to keep the rates up. Last year, for example, I made over 95,000 connections with over 96% file integrity. It was great - almost broke the 100K barrier. But it looked like the conditions this year would hold down the totals to 80K or less, despite the newer software and better net access available.

As I cruised the nodes on the lower-speed subnets looking for multipliers, the rate slowed down a bit. Found a full node in East Carribea, and picked up all the easy ones down there. Even found the Bonaire Expedition team there - those guys are a guaranteed 6-node connection every year. And this time, they asked me to go to two additional nodes. I was glad to oblige, since this sector has gotten rare. Most serious competitors have relocated to physical coordinates with better net access. I'm one of the few serious ops left in NE01, struggling with our poor net access...

As time wore on, I noticed that I was actually ahead of the connection total I had at the same point last year. The rate meter was hanging up there at a brisk 3500/hour pace. I popped the auto-scan routines up, and settled for a short nap so I'd be ready for the net speed peaks around sunrise in Europia. I had all the low-speed subnet nodes lined up for my announce burst, which would be inserted in the high-speed queue right after the enviro-report.

While I slept, the machines relentlessly scanned the low-speed subnets, trying all loggable nodes, and posting announcements of when we'd be back for more connections. This trick was a little mod I'd installed in the stock Microsoft Contest software. Like all mods, it had to be approved by the Contest Software Council. This was a consequence of the time a few years back when some newbie popped a loop into one of the calling routines and got stuck in a sending sequence on some image-transmission subnet. Those image guys hate it when a contester lands on *their* net. Unauthorized mods are now illegal. I sort of liked it the old- fashioned way, since now the nets are crowded with these appliance ops who just use the software as it comes. If they have good access and a hot machine, they can sometimes post pretty good scores. But I like it when I out-operate these guys with my lousy access and old machines by out- operating them with my own software tricks.

When I woke up, I knew immediately there was a problem. The screen on the left machine showed a smoke icon on it. The machine on the right was still OK, so I tapped a few keys to acknowledge human presence, let it run, and set about fixing the left machine. After a few minutes of diagnostic routines, I found the problem. The machine was OK. It had been fooled into thinking it was out of sync when it began a connection with a station on one high-speed net, and found the machine on the right was connecting to the same station on another high-speed net! Since simul-connects were no longer permitted (the rules clearly state that only one symbol may be transmitted in any one 100 nanosecond period unless the second transmission is a new multiplier and must be connected at a lower speed), the machine concluded that the internal clocks had gone out of sync. Stuff like this happens when guys bend the rules on the other end. I documented the incident and sent it off to the Judges on the spotting machine, including the time/net/file and timing traceability. I also asked for and received permission to disable the timing-sync-hang feature, since there were probably other rule-benders on the net. All this only took a few minutes, but the rate was down in the 800-900 range, so I didn't lose too much time. Good thing this didn't happen at peak rate time! And the offending simul-connect guy will be warned. If he does it again, he'll go to jail.

Next the spotting machine came to life with a live feed from the Moscopolis main node. The enviro-report was just wrapping up. Bang! There I was, with my solicitation announcement, with protocol and modes in the clear before all hell broke loose. A perfect tail-end, just like the old days! The screen went wild with file exchanges. Sometimes the subnet would jam, and I'd have to hop to another - fortunately I had a list of machine-level node addresses so I could transfer to another switch in native mode, not the slower top-level mode that the newbies used. I always got there first. A big part of the technique in this game was knowing when to move to another node and when to keep on the same one. As long as you had primary status and the node wasn't too heavily multiplexed, you could cruise along.

As the sun rose across Europia, I had my nodes all staked out. The same scenario was played out over and over. In some time zones, my tailend announced a later time and port for connections, since I couldn't handle more than two streams with my limited access. Knowing which ones would have the right combination of activity and low congestion was the way to go. I always went for node in the higher latitudes first, since the lower latitude dwellers would likely be indoors most of the day due to the solar radiation dangers. I could get those guys later when the traffic thinned out. But there were always the macho guys who wanted to hold primary status on these nodes as long as they could, using their high-speed access ports and soliciting continuously. I prefer the nets with clearer channels, and sometimes set myself with a higher position in the user list to attract the casual operator who doesn't want to slog through the re-tries at the node-edge (as we call the primary user ports).

Things went well all day long. As I'd guessed, there was enough activity in Northern Europia to keep the rate up with hardly any re-tries. Lots of good sectors called in, including some night-owl Asians through some kind of worm-hole path. Openings like that are really cool. The rate meter stayed well over 4000 for most of the shared daylight hours. The local access switch here in NE01 has pretty reliable links to the Euronet most of the time. As long as I get there first, I can usually win the contest with this opening.

As darkness fell across Europia, I sent the left machine to scan the Africano network. Still not much rate available there, but there sure are a lot of multipliers. I called up one of my routines to look for low-density traffic, and found a group of missionaries chatting. Several rare sectors were there, and I multiplexed into their stream. After a few seconds to make sure I hadn't been detected by any of my competition, I announced my presence and requested file exchanges. The missionaries were cooperative, and asked what kind of files I needed for a valid exchange. I explained the requirements, and one by one, they began connecting and transferring data to me. One guy sent a file too small to qualify as an exchange, so I had him send it twice to make the minimum size. Then I asked the whole group for connections on another node, then another, and another. By the time I got to the fourth node, most of them had run out of files (after all, these guys were just casual ops), and the competition had found them and the nodes were getting crowded. But I had scored big time here, and thanked all of them actually, I sent a broadcast thank-you to all connected stations in several protocols, included the inflated one, just to slow down the other guys...), and I was on my way.

A quick spin of the high-speed pipes to Carribia and Oceania, and I had a full slate of multipliers. Then it was back to the Asian nodes as they came back to life. I hit the same node I had plundered the first night, and ran it dry. I hopped around the other nodes in the sector, trying to maximize the rate while maintaining file integrity.

When I finally looked over at the spotting machine, I saw that I was ahead of the pack after the first 24 hours. I only had a few thousand more connections (at 63,000) and about a hundred more multipliers more than the next guy. Modest leads like that can evaporate quickly, but my file integrity spec was at 99.8%, while nobody else was over 96%! I guess all that time I spent at the machine-code level optimizing the error- retry-timing loops for compressed-mode files was worth it. Most guys just give up and accept a corrupted file and let the Judges worry about it. Not me. Not since I lost the '02 contest by having 500 connections thrown out for checksum violations in the decompressed files. Back then, I threw out the real-time checking and let the decompressor run at the submission level like most guys. I just submitted whatever the machine copied. And I paid for it big time that year. Some guys have been known to decompress all files after the contest and check for errors after the contest is over. One guy even interleaved the checking/correction process with a low- rate submission! As long as he could locate the errors before his submission stream got to that point in the log, he could fix it. His integrity rates were always pretty high, but everyone knew what he was doing. Eventually they'll make a rule about that. Anyway, my submission is at high-speed, so everyone knows I'm getting the files right in real-connection-time.

So on it went. The rate fell predictably to sub-1000 levels, since I'd already made connections with most of the serious guys, and now it was time to seek out the casual guys. Had to stoop to surfing some of the newbie-populated nets, but even that got thin after a while. Then I kicked in my latest trick - the auto-pass. I scanned the log to identify which stations I needed on other nodes. The system then launched probes to the right nodes to tell those stations that I needed some node-connects from them, and embedded files with the connection protocol for the needed nodes with the probe. All they had to do was acknowledge and their machine would auto-reroute a real- time return, and the file would come back to me, completing a new node-connection! It was almost like those QSL cards in the old days, where the card was sent already filled out, and all the recipient had to do was sign and mail it. This trick will eventually get copied, too. But I'm the first to use it - sort of like the first guy who asked a DX station to QSY to another band for another QSO. Soon it'll be common practice. And I'll have to think up another trick.

I set the machines on autopilot while I grabbed another nap. Through the night, the machines searched for new stations, received return connections from the auto-probes, and automatically requested multiple node- connects from each station that replied. Sure glad the machines don't get bored. I know I would.

I woke up to a pleasant surprise. The WWV reports on the spotting machine showed that the Europian agros had outvoted the urbans, and there would be rain all day in most of Europia. That usually means more people indoors all day, and more connections available. There was also info from the air travel reservation system that told me that several of the top DXpeditioners from the Northern Europia sectors like EU92 and EU93 (what used to be Finland) had boarded flights to some of the Atlantian and West Africano sectors where there hadn't been any activity yet. My guess was that they were planning entries in the 6-hour DXpedition class. The pileups would jam the nets, but I've figured out how to beat the crowd. I had auto-probes lurking in the passport- validation streams of all the landing ports for these guys' flights. As soon as they got off the plane and their machines received the Immigration validation file, my probe snagged them for a quick 1-file connection and exchange. By the time they got settled in to get serious, they found I'd already contacted them. If I could compress by another factor of 10, I could have re-routed the connects to a few more nodes, but I had I was happy to get the sector on one node anyway. Lots of other guys wouldn't be able to break the collisions and get even one file through with no errors.

The rate to Europia didn't let me down - it was just about as good as the first day. The rate meter was hanging in the low 3000s, and it looked like I would make it to about 95K. I got a note on the spotting machine that says the Infotainment programs on the cable were already predicting me as the winner, with six hours left to go, and I expected a reporter to drop by for an interview.

After darkness fell in Europia, the rate fell precipitously, as people left interactive mode and went into passive mode on the 'Tainment wires. With rates falling to 500/hour, I sent the machines on a final search-andconnect tour of the low-speed subnets, occasionally stopping by the main Web backbone to watch the traffic jams on the DXpeditioners. A couple of them were using some kind of auto-select to thin out the mess, and were actually getting first- try complete files with no collisions. You can always tell the good ops by the way they tailor their software to handle the pileups. Some of the pileups on the first-timers were pretty bad, though. You could almost see the newbies clicking over and over again on the auto-repeat icon on the MS Contest S&P menu, even when the DX guy was connected and sending. Others have adopted that annoying habit of only signing the last two header strings instead of the full header. They just don't understand how much that slows everyone down.

With an hour to go, I got that rush of adrenaline again. I took the machines out of auto mode, and took the controls live myself. I think only Fred Laun ([email protected]) and I still do this. My fingers flew over the keyboards as I scanned the nodes, watching the screens for ragments of ID strings in the murk. My concentration sharpened as a handful of new stations appeared, and I was able to connect, exchange, and move to another node faster than the neuralware recognizer could fill in the missing pieces of the headers. These machines were good, but even their fastest rules-based algorithms were still no match for the human inference machine between my ears. The rate meter actually began to climb as I launched auto-probes to some of the Pan-Pacific satellite nodes, and began to get replies and node-hop requests. The spotting machine beeped, and when I looked over, I saw a note that I'd have to supply a urine specimen for stimulant-screening before my score would be accepted. When I went to full-manual mode and the rate went up the machine detected it. The machine figures any rate over 1500 in manual mode must be chemically-enhanced, and I've entered in the unenhanced category. This flag is kind of a nuisance, because I've done over 2500 in manual mode several times, but I can't shut the warning off or I'll get hit with disqualification and a 5,000 credit fine for the first offense. I stood up, walked over to the spotting machine, and delivered the required specimen in real-time to the analyzer port. Any delay would have been interpreted as fraud. Then it was back to the 'boards for another 15 minutes of furious connecting.

When it was over my preliminary final score was 95,315 connections and 2560 sector- nodes, for a final score of just over 683 Meg. Once the files were scanned for accuracy, that'd come down a bit, but the raw integrity rating looked to be in the 99.7% range so I didn't expect to lose much. A bit short of the record from last year, but darned good considering the conditions. The claimed scores jumped onto the screen as all the major competitors accessed the so- called "3830 page". My machine logged them all in, and ordered them, with raw and projected final scores. A few minutes later, the doorbell rang, and the reporter was there, as expected. By the time I showed her to the shack, the scores had been validated. I had won.

Just for kicks, I sent the spotting machine to the 'Tainment net, and downloaded the whole interview. It was the usual stuff - the reporter played back some of the screens they had observed at Network Central Control and asked a few questions about how I thought up the software hacks. When the interview went interactive, all the usual questions were there from the first-timers: "How did you learn to program the real-time decompressor?"; "How did you know when it was time to jump nodes?"; "How can I compete with guys like you when I only have one machine and a lousy 64kb/s access port?"; "Don't you think there should be a 24-hour category for those of us who aren't able to go the full 48?" and all the rest. I patiently responded that most of the skill is developed over years of experience, and I didn't win my first contest either. Then I pointed out that my own equipment is pretty modest, and operating skill and knowledge of the code (the software code, that is) can overcome a lot of hardware shortcomings. But as usual, most people were tuning out by then, since most people don't want to hear old-timers like me talk about using hard work to achieve success.

Hopefully, a few of them were still watching when the reporter handed me the ceremonial check for 10,000 credits. Then the spotting machine task-switched to personal account mode, and reported that my account was at that exact second acknowledging receipt of 10,000 credits. And within seconds, several screens appeared from IBMwood, SonYaesu, and others soliciting me to buy their new hot boxes. But I think I'll rest until the next contest and use the same old machines. I've grown attached to them, and I'm afraid I won't have the feel of any new machine before next weekend's Sprint. That event is so short, and accuracy is so crucial that you just can't compete with an unfamiliar machine. But that new IBMwood with the Intelrola Hexadecium(tm) processor array sure would make my decompressor sing...

See you on the wire

Member Comments: Add A Comment