Cory Hamasaki's DC Y2K Weather Report V2, # 31 "July 30, 1998 - 529 days to go." WRP87 (c) 1997, 1998 Cory Hamasaki - I grant permission to distribute and reproduce this newsletter as long as this entire document is reproduced in its entirety. You may optionally quote an individual article but you should include this header down to the tearline. I do not grant permission to a commercial publisher to reprint this in print media. As seen in USENET:comp.software.year-2000 http://www.elmbronze.demon.co.uk/year2000/ http://www.kiyoinc.com/HHResCo.html Don't forget, the Y2K chat-line: http://www.ntplx.net/~rgearity any evening, 8-10PM EST. --------------------tearline ----------------------------- Please fax or email copies of this to your geek pals, especially those idiots who keep sending you lightbulb, blonde, or Bill Gates jokes, and urban legends like the Arizona rocket car story. If you have a Y2K webpage, feel free to host the Weather Reports. Did you miss Geek Out? Project Dumbass needs you. In this issue: 1. Warn the Geezers. 2. True Facts 3. Don't Fix until Failure 4. Citizens for Y2K Recovery 5. DISASM 6. The Jo Anne Effect 7. BT's Funeral 8. CCCC ------- Warn the Geezers ----------------- Last weekend a geekette pal saw the POS at Trak Auto fail. No big deal right? How many times have you seen a store computer fail in the last 20 years? I've never seen it happen. Well, today I was in the Giant Grocery store, buying my day-old donuts, ...a frozen pie, the new extra large size potato chips, a couple 2 liter jugs of house brand soda, what we geeks call haute cuisine, and guess what, the geezer in front of me tried to pay with a credit card.... IGG0XYZI - Communications Timeout. The clerk went into rambling mode about bank problems getting worse ... you listening Arnold? ... how it's happening a couple times a week now... and the geezer ... lord he was a natterer, one of those nobody-home types, didn't know what to do, the clerk instructed him to swipe his card again ... which he did after looking around a couple times... what was the point of looking around... who knows but geezers do that ... it's really a problem when they're driving on the beltway and something unusual happens, like a big truck goes by, then they look around, watch-out, watch-out, watch-out. -crash- ... the leader of the pack... oops, free associating again... did I forget to mention that Dolly Parton was on the cover of one of the tabloids... check out the story. I was in line for 10 minutes while the geezer swiped his card, the clerk talked about the worsening communications problems, and finally they opened the base of the conveyor belt and reset the machine. I told them, "have cash, in about a year you will be on a cash only basis." The clerk got it, oh yeah, the computer problems are getting worse... but the geezer was even more confused than before. He did more looking around and I could tell he wasn't tracking me. "Have cash, cash is good." The geezer blinked his eyes... nobody-home. Anyone else out there notice POS or credit card system failures? ------- True Facts ----------------- Here are the known facts: 1. Most systems have a date flaw in them. Large complex systems certainly have a problem. Large enterprises depend upon multiple large complex systems. 2. Some enterprises started early, years ago; spent a fortune, alternately whipped their geeks and kissed them; they're not done. These include BankBoston, Fidelity Investments... I know of others. They might make it, then again, they might not. Do you feel lucky today... well do you, punk? 3. Most enterprises didn't start until sometime this year. GM, BankAmerica may fall in this category. It is unlikely that anyone starting in 1998 will complete remediation, testing, in time. 4. Some enterprises are yet to start, the U.S. IRS and DoD may fall into this category. Maybe, maybe not but the DoD's Inspector General raked the DoD over the coals for falsifying compliance reports. This is what we in the c.s.y2k call a bad sign. You may interpret it otherwise... most of us are not encouraged by stunts like that. 5. Fix on failure doesn't work. If you could fix a mission critical enterprise wide system in the first 2 months of 2000, fix it before your creditor's lawyers tear you to pieces, why don't you fix it now... this weekend, this month, and slide into Y2K smooth... like a masamune katana into a torso. 6. We've had indepth discussions on specific areas, Telcos, Rail shipment, ICBM launch controls, Elevators, Nuclear Power Plants, Unix. Invariably, the initial discussion is back and forth, with the "No Problems Here" crowd submitting convincing arguments. In all cases (except maybe Firetrucks), the debate has fizzled out when specific, unsolved flaws are reported. A good example is Telco's. We're in the last stages of the Telco debate, the "No Problems Here" crowd are now reporting that Telco Switch vendors have *announced* products with Y2K fixes, due in 3-6 months... any day.... real soon... trust us.... Based on the evidence.... nothing will work. We doomed.... run, run Dogbert, run for your life. ...but optimistic FOOL that *I* am, I'm hoping that something will be done in the next 520 days. Summary - At this point, there is *no* debate. The systems will fail and civilization will collapse... The only question on the floor is: Will enough be done to fix enough of the essential systems to prevent complete collapse? Oh, by the way, Al Gore says that the Federal Government will be done by March 1999. Flash, Ray Long says the FAA will be done September 1998, done, complete, just some testing left. ------- Don't Fix Until Failure ------------- Instead of Y2K'ing, a very large computer services vendor is kicking back, sure, they'll do some work for a select group of loyal old clients but new clients? Nah, they're.... as they say, fully engaged. Then in 520 days when the failures occur and the clients start screaming "We're doomed! We're doomed!" This very large computer services vendor will stroll in and help out... but not for $100/hour... not for $200/hour, for significantly more. They'll be calm, professional, and will have "take it or leave it" contracts, contracts with large boxes for the amounts. This strategy gets solves the legal liability issues because ... the systems will already be broken. ...this is a strong reason that the work is *not* being done now... why fix code for $100/hour when you can wait a few months and do the same work for much, much more. Wait a few months and not have any liability. So the smart players are sitting this game out, liability, low rates, sit this game out and let the lightweights debate the morality of scaremongering... Let the nubies, who don't even know what the game is, have their day, and in 520 days, they'll bring in the big-guns. In 520 days, they'll say, you're mine, punk. ....sign over the firm. --------- Citizens for Y2K Recovery ----------- Finally someone is doing something.... At the last WDC Y2K, I spoke with the reps from "Citizens for Y2K Recovery". These are people who are not debating the problem, not running for cover, not fixing code... they're prepping their communities for survival and recovery. They asked me to speak at their conference September 26th. Others will be there, Jim Lord, Tony Keys, Rick Cowles, Bruce Webster and more. A full line up of the concerned and hopefully less confused than me. Their webpage is at http://www.y2krecovery.org The issue isn't will the systems fail, they will. The issue is preparing for the event and making the most of the next 520 days. I've signed up to talk about emergency communications. ----------- DISASM -------------------------- I have a copy of an S/370 disassember. It's written in S/370 assembly language. I haven't piped it over to the mainframe yet. I also have a copy of the public domain S/370 emulator and assembler that runs on Intel DOS. I was thinking of compiling the emulator in Mark Williams C. I have Borland's Turbo C++ for Windows and IBM's CSet++ 2.1. Mark Williams is a complete DOS C and should be adequate for the emulator. The disassembler, emulator, and assembler would make a nice S/370 fixit kit. Here's some text for the disassembler: ****************************************************************** 07690002 * * 07700002 * THIS ROUTINE FORMATS RR-FORMAT INSTRUCTIONS FOR OUTPUT. * 07710002 * * 07720002 ****************************************************************** 07730002 * 07740002 RROPND EQU * *** RR FORMAT INSTRUCTIONS *** 07750002 CLI 0(R7),X'05' IS IT BALR 07760002 BNE RRSTRT 07770002 CLI 1(R7),X'EF' IS IT BALR 14,15 07780002 BE RRSTDL YES 07790002 TM 1(R7),X'0F' NO, IS R2 = 0 07800002 BNZ RRNSTD NO 07810002 MVC COMMENT(11),=C'ADDRESS SET' 07820002 B RRSTRT CONTINUE 07830002 RRSTDL MVC COMMENT(11),=C'STD LINKAGE' 07840002 B RRSTRT CONTINUE 07850002 RRNSTD MVC COMMENT(14),=C'NONSTD LINKAGE' 07860002 RRSTRT EQU * 07870002 LA R15,TEXT+1 POINT TO TEXT 07880002 LR R11,R1 COPY INSTRUCTION TYPE 07890002 CH R11,=H'48' RRE TYPE? 07900002 BL *+8 NO, JUST RR 07910002 LA R15,2(R15) ADJ FO 2 BYTE OPCODE 07920002 SR R1,R1 CLEAR WORK 07930002 IC R1,0(R15) GET 2ND BYTE 07940002 SRL R1,4 SHIFT OUT R2 07950002 CVD R1,DBLWD CONVERT 07960002 OI DBLWD+7,X'0F' CLEAR SIGN 07970002 UNPK DBLWD(2),DBLWD+6(2) UNPACK 07980002 MVI OPNDS,C'R' BEGIN R1 OPERAND 07990002 CH R1,H10 REG NBR > 9 08000002 BL RR11 NO, 0-9 08010002 MVC OPNDS+1(2),DBLWD YES, 10-15 08020002 LA R1,OPNDS+3 TO NEXT POS 08030002 B RRCMA CONTINUE 08040002 RR11 MVC OPNDS+1(1),DBLWD+1 MOVE REG NBR 08050002 LA R1,OPNDS+2 TO NEXT POS 08060002 RRCMA EQU * 08070002 CH R11,=H'48' RRE WITH ONE OPERAND? 08080002 BE RRXIT YES 08090002 CLI TEXT,X'04' IS IT SPM 08100002 BNE RRCMA1 NO 08110002 TM 1(R7),X'0F' YES, IS R2 FIELD = 0 08120002 BZ RRXIT YES, GOOD INSTR 08130002 B 4(R9) ERROR RETURN (NOT INSTR) 08140002 RRCMA1 MVC 0(2,R1),=C',R' DELIMITERS 08150002 PACK DBLWD(1),0(1,R15) FLIP 2ND BYTE 08160002 SR R15,R15 CLEAR WORK 08170002 IC R15,DBLWD GET R2R1 08180002 SRL R15,4 SHIFT OUT R1 08190002 CVD R15,DBLWD CONVERT 08200002 OI DBLWD+7,X'0F' CLEAR SIGN 08210002 UNPK DBLWD(2),DBLWD+6(2) UNPACK 08220002 CH R15,H10 R2 > 9 08230002 BL RR21 NO, 0-9 08240002 MVC 2(2,R1),DBLWD YES, MOVE REG 10-15 08250002 BR R9 EXIT 08260002 RR21 MVC 2(1,R1),DBLWD+1 MOVE REG 0-9 08270002 RRXIT BR R9 EXIT 08280002 I think it's public domain source. What's neat about a disassembler is that it's almost the same as an emulator. It has to understand the format of the binary, opcodes, operands, and data. What's differendt is that the disassembler generates source code while an emulator makes state changes in the psuedo machine. There's been a discussion on c.s.y2k on disassemblers, Arnold did a hand disassembly on a module. This is a tiny part of a program that does it mechanically. Please denial-heads, feel free to explain the above code fragment. Tell us about how your big W95 using brain has it all figured out, ------------- the Jo Anne Effect -------------- I'm expecting the failures to take this pattern. This graph is not to scale, it's the shape that counts. jul X <---<<< we're here aug X sep XX oct XX nov XXX dec XXXXX 1999 The Jo Anne effect. jan XXXXXXXX feb XXXXX mar XXX apr XXX may XXX jun XXX jul XXX aug XXXX GPS sep XXXXXXX fiscal year stuff. oct XXXX nov XXXXX dec XXXXXXXXXX first 2000 transactions. 2000 jan XXXXXXXXXXXXXX feb XXXXXXXXXXX mar XXXXXXXXX apr XXXXXXXXX may XXXXXXXXXX jun XXXXXXXXX jul XXXXXXXX aug XXXXXX sep XXXXXX oct XXXX nov XXXXXX dec XXXXXXXXXXX 2001 jan XXXXXXXXXXXXXXX lookbacks feb XXXXXXXXXXX mar XXXXXXXXX The Jo Anne effect is caused by systems that attempt to sort data into last year, this year, and next year piles. This is different from lookaheads and predictive systems. This is strictly a wild guess but I think the Jo Anne effect will be less than December 31, 1999 failures. ------------ BT's Funeral ------- It was well done... BT had retired from the Navy and his pallbarers were dressed in Navy whites. I stood respectfully to the side and took pictures, maybe I can get them scanned in. The chaplain spoke, the honor guard saluted and folded the flag, there was a 21 gun salute, they rang the chapel bells, the lone bugler played taps... BT's family and geek friends said their goodbyes. It was a hot, sad, DC summer day in Arlington National Cemetery. I was the last one to leave the grave site. BT was working on a rehost for a mission critical system, trying to beat Y2K but not working on the core system that he, Shmuel and I worked on in the 1980's. This was a side sub-system. It will be that much harder without one less pair of hands at the keyboard. One less geek who knows PL/I, C/C++, and S/370 assembler. BT had unusual application specific knowledge, he was a domain expert who could program. There are very few people in the world who knew what he knew... perhaps two or three. ------------ CCCC --------------- More clueless than normal... that's me. 520 days, it's going bad... 20 days until the 500 day decision point. The magic that keeps the world's population fed and the life support systems going is about to run out. There were large cities in ancient times, but what were the numbers... what was the population of Imperial Rome, Socrates' Athens, Tikal, 17th century London? I've been to Tikal, seen the ruins, climbed to the top of the temple of the Giant Jaguar, seen the reconstructions, what happened? How fast did it go bad? Was it disease? War? or was it something more ominous, the pendulum swung back and took the civilization with it. Suppose it's not personal, suppose it's just chance, just the luck of the draw and our luck is running out. Random chance and the power of cycles and we're looking at 20 years of a decreasing standard of living. Decreasing or a sudden drop in 520 days and a stair step down from that as each month passes. That's the model for sepsis... you get a little cut, your normal skin flora invade and overwhelm your defenses but just at that one spot; the debris and toxic byproducts of the localized battle kill more cells and that leads to more debris. It's like a fire eating its way out from that spot, septic shock sets in, your heart pounds itself to death. Look at you. You're a system of systems, layers interlocking, complex, signals, electro, chemical, self regulating and self repairing... one failure can be bypassed, a dozen might be noticed, but thousands and thousands of system failures, a vitamin deficiency, overwork, too much fat, a pneumococcus, an endocrine imbalance, fatigue, lots and lots of things just a little out of spec and you die. In 520 days, civilization will be out of spec in millions of ways. What's the solution? Fix code like a maniac? Sure, if they'll let you... I know geeks, real power-geeks, who have been ordered not to make Y2K fixes. ... replacement systems are in the wings, they'll be ready real soon now, trust us... Remediation is running on less than half power, it's not a problem, let's have a meeting, look out the window... hey, is that the new Porsche? They're not working on the code. It's business as usual in 75% of the shops... what's this Y2K anyway. Run for the hills? Maybe... can you outrun a 1 megaton airburst... how about something more scary... a nice resistant staph ... more soldiers die from disease than bullets. Maybe if you have an isolated place, clean water, and lots of fresh air... Organize your community, pre position fuel, water purification supplies, food, have a real plan to include specifics, regular SETs (simulated emergency tests) When I say plan, I don't mean pop-biz-jive unrelated to reality, metrics, mission statements... I mean hard specifics. September - Purchase n-tons of emergency food. October - Generators, fuel tanks. November - back up communications, CB and Ham radio. Hard specifics, supplies in place, a plan to disconnect the community from the grid, take care of the retirees, the children, make it work. scale back, simpler, living within bounds... fresh produce in season and dried or canned out of season. What's really going on? The FAA won't make September 1998, Ray Long was telling a joke. The Feds won't make March 1999, koshecan, gore, what kidders you are. The SIA Street test was a failure, the news blackout is the first clue... the whispers from insiders in NY is the second, ...check back through the old DC Y2K WRPs, the superprogrammer's grapevine hasn't missed yet. Read Dilbert, geek to geek, speak in the code of the geeks, what's really going on? The NYSE is about to take a hit in the machine check handler and HIR will be UNsuccessful. The message is, "HIR FAILED, SYSTEM ENTERING WAIT STATE." cory hamasaki 519 days now...