Author: Dann Corbit
Date: 17:45:00 11/17/04
Go up one level in this thread
On November 17, 2004 at 20:40:02, Dann Corbit wrote: >On November 17, 2004 at 20:21:27, Uri Blass wrote: > >>On November 17, 2004 at 20:03:39, Dann Corbit wrote: >> >>>On November 17, 2004 at 18:56:13, Uri Blass wrote: >>> >>>>On November 17, 2004 at 18:33:11, Dann Corbit wrote: >>>> >>>>>On November 17, 2004 at 18:00:38, Uri Blass wrote: >>>>> >>>>>>On November 17, 2004 at 17:16:37, Dann Corbit wrote: >>>>>> >>>>>>>On November 17, 2004 at 00:33:30, Uri Blass wrote: >>>>>>> >>>>>>>>On November 17, 2004 at 00:22:01, Uri Blass wrote: >>>>>>>> >>>>>>>>>On November 16, 2004 at 20:14:23, Jonas Bylund wrote: >>>>>>>>> >>>>>>>>>>On November 16, 2004 at 19:49:37, Dick Long wrote: >>>>>>>>>> >>>>>>>>>>>On November 16, 2004 at 19:32:57, Mike Byrne wrote: >>>>>>>>>>> >>>>>>>>>>>>On November 16, 2004 at 19:13:03, Dick Long wrote: >>>>>>>>>>>> >>>>>>>>>>>>>Did Ferret ever become Freeware? If so where can you get it? Just wondering >>>>>>>>>>>>>because Bruce after years of promise never has and obviously never will put it >>>>>>>>>>>>>on the market. Further it's not as good compared to other programs now as it was >>>>>>>>>>>>>vs some older programs. >>>>>>>>>>>>> >>>>>>>>>>>>>Thanks >>>>>>>>>>>> >>>>>>>>>>>> >>>>>>>>>>>>No - but it was clearly near the top at one time and at it's best , it was top >>>>>>>>>>>>amatuer for its day. >>>>>>>>>>>> >>>>>>>>>>>>Perhaps some day he will release it - but I have no reason to suspect that he >>>>>>>>>>>>may. >>>>>>>>>>> >>>>>>>>>>>Thanks , Too bad years ago he could have sold it to rebel or chessbase, and made >>>>>>>>>>>a nice piece of change easy. He just kept hmmming and hawing, i was like come >>>>>>>>>>>on bruce. He just refused. Tsk Tsk Tsk. >>>>>>>>>> >>>>>>>>>> >>>>>>>>>>Not everbody is interested in money... >>>>>>>>> >>>>>>>>>Of course but Bruce also never made it a free program and never sent it to >>>>>>>>>tournaments like Leo's WBEC so everybody could see what is it's strength >>>>>>>>>relative to other programs on equal hardware. >>>>>>>>> >>>>>>>>>Uri >>>>>>>> >>>>>>>>Of course Bruce does not have to explain and it is his own program >>>>>>>>but the fact that he never sent it to tournaments like WBEC or other tournaments >>>>>>>>is something that I do not understand because I expect programmers to be >>>>>>>>interested in results of their own program against different programs. >>>>>>>> >>>>>>>>Not releasing a free program is clearly understandable because of all the clones >>>>>>>>and I can also understand a decision not to sell the program if you believe that >>>>>>>>you cannot make enough money from it(I have no idea how much money earn the >>>>>>>>programmers of ktulu or partiot and maybe Bruce believes that the money that he >>>>>>>>can earn or could earn from selling Ferret is not enough money to justify caring >>>>>>>>about customers). >>>>>>> >>>>>>>I can think of lots of reasons why a person will not want to go commercial. >>>>>>> >>>>>>>1. You may lose a great deal of money. How possible? >>>>>>>Suppose that you make a royalty of $2 per copy of the program sold, and 200,000 >>>>>>>are sold. Sounds pretty good, because that is $400,000. But if in the same >>>>>>>time span you spend 10,000 hours on bug fixing, enhancements, tech support, etc, >>>>>>>then you made $40/hour. Sound pretty good? A programmer like Bruce can >>>>>>>definitely make $100/hour, so he lost $60/hour. Multiply by 10,000 hours and it >>>>>>>is a pretty good chunk of change. >>>>>>> >>>>>>>2. When you go professional, chess programming will cease to become a hobby and >>>>>>>instead become a job. >>>>>>>It's like the difference between digging a ditch and weight lifting. Nobody >>>>>>>wants to dig a ditch. But everyone loves to lift weights. Counter-intuitive, >>>>>>>but a real psychological phenomenon. When something ceases to become recreation >>>>>>>and instead becomes a job, all the fun can go out of it. >>>>>>> >>>>>>>3. There are dangers associated with any sort of chess programming. Suppose >>>>>>>you have a bug in your hobby program that it leaks memory. Who cares? It's a >>>>>>>hobby program. But if it is commercial then you _HAVE_ to fix it, and as soon >>>>>>>as possible or you are negligent. >>>>>>> >>>>>>>4. If you go commercial, then you have a responsibility to maintain the >>>>>>>product. If you release it and it needs corrections, it is not a friendly thing >>>>>>>to do to just try to walk away from it. How long will you be tied to >>>>>>>maintenance? Over the lifespan of any software product, 80% of the cost is in >>>>>>>maintenance. >>>>>> >>>>>>I do not believe that all the programs are the same. >>>>>>Maybe it is correct for programs in other fields but chess programs are >>>>>>different. >>>>> >>>>>Chess programs have no exemption. The 80/20 rule for software maintenance has >>>>>been known for decades. If you do not plan for it, then it is an absolute sure >>>>>disaster that will definitely happen. >>>> >>>>I know that programmers like Amir Ban have full time job not in chess >>>>programming and I did not see disaster. >>> >>>He wrote his program many years ago. In November 1997, he won the world >>>champion, still as an amateur. If he honestly tells you the hours he has spent >>>on it before then and since then, I am sure that the 80/20 rule applies. >> >>I do not know because I did not ask him but Junior of 1997 was clearly weaker >>than Junior of today and I guess that he spent hours on improving the engine. >> >>I believe that he spent thousands of hours before 1997. >> >>If we assume that he spent 5000 hours before 1997 then to keep the 80:20 he >>needed to spend 20000 hours about mintanence even without improving the engine. > >Maintenance includes improvements (of course). In fact, every act of >maintenance is also an act of improvment, if you think about it. (Bug >corrections, documentation, etc.). > >>20,000 hours in 10 years are more than 5 hours per day and he has full time job >>so he could have no time for improving the engine. > >52 work weeks per year * 5 days per work week * 10 years = 26000 work days. >That figures out to less than one hour per day. I'll save you some time -- it is 2600 days (at 8 hours per day a maximum of 20,800 could be spent). So I guess he spent 5000 hours. Which would put development time at 1000 hours. I would be interested to know from Mr. Ban how accurate those figures may be. >>>>>> Chess programs represent thousands of hours of work for the very >>>>>>>top performers like Shredder or Ferret. Which means that the maintenance work >>>>>>>will be tens of thousands of hours. >>>>>> >>>>>>I see no reason that maintenance work will be tens of thousands of hours for >>>>>>chess programs. >>>>> >>>>>It is not just time spent sitting down at the keyboard typing code and testing. >>>>>It is time spent documenting. It is time spent talking about new features. It >>>>>is time spent in meetings that are tangential to the tasks at hand. It is the >>>>>total cost of maintaining a program. >>>> >>>>Nice theory but even sitting down and writing code to fix bugs that users do not >>>>like is not done and Amir does not care about cases when Junior cannot find >>>>simple things if it has not big influence on it's rating. >>> >>>In any case, he will have spent far more time in maintenance than in writing it. >>> >>>>>>It seems to me that for chess programs maintenance of commercial program is >>>>>>clearly less time then developing the program. >>>>> >>>>>Not a chance. >>>>>Unless there is some magic wand that can remove the burdens of software >>>>>maintenance from a chess program, it will be exactly like all the others. >>>> >>>>Yes >>>>There is some magic that is simply not caring about it. >>> >>>That is called negligence. If you are told about it and do not act, it becomes >>>malice. Malice can pierce the veil of corporate invulnerability and land you in >>>jail. >> >>I do not know about people who went to jail only because of not fixing bugs in >>software that they sold. > >People have been killed by faulty software (e.g. an X-ray machine with 10K times >the requested dosage). Billion dollar losses have happened due to software >defects (The Arianne rocket failure was due to data type conversion error). > >If there is a defect and you know about it, and it can cause a loss, then you >have to fix it. "Have less fun" is not a loss, but "Crash a computer during >business operations" is a loss. > >The Tacoma Narrows bridge had defects in materials and other problems. There >were a few acts of negligence. But if the people who did it KNEW what was wrong >and took no action, can you imagine the repercussions? > >Some useful links: >Software quality: >http://home.flash.net/~lyttlec/SE/se.html >http://csdl.computer.org/comp/mags/so/1996/01/s1012abs.htm > >Y2K lawsuits (a web search will turn up thousands of links): >http://www.avaya.com/ac/common/index.jhtml?location=M1H2G5F5017&rec_id=pr-040514a-1002104427 >http://static.highbeam.com/b/banktechnologynews/july011998/interactivebeatintuithitbyy2klawsuit/ > >Look for "Denver Airport Software" on Google. Also good are: FAA's Air Traffic >Control System; American Airlines' Confirm; and Bank of America's MasterNet.
This page took 0 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.