The new credit system xxxxx.

I'm sorry but a 40 point dif, on the same time w/u's is just plain wrong.

Whats wrong with this credit system. A lot. There is no way that the same box should have a +/- 40 differential on 8 hr wu's.

https://boinc.bakerlab.org/rosetta/results.php?hostid=263683

Go ahead and explain to me how this is so much better than the old system.... Where I was getting just about the same 1500 ppd as I was on Predictor. And I have added 145mhz of overclock since Predictor went down.

I've reached my 200 point per day decrease from what I was getting on Predictor with a quorum of 3. Something smells and its the new credit system. So much for the cross platform uniformity. If this does not change I will have to pull these boxes. Its just NOT fair. Period.

---

BETA = Bahhh

Way too many errors, killing both the credit & RAC.

And I still think the (New and Improved) credit system is not ready for prime time...

Whats wrong with this credit system. A lot. There is no way that the same box should have a +/- 40 differential on 8 hr wu's.

I agree that 40 points +/- is to much.

But if you look at your computers they seem to get a bit more than they

claim at averige.

Anders n

---

I don't know exactly how the credit is calculated, but when I look at your recent results, it's clear that you have some cases where you get more and some cases where you get less than your claim. In the cases where you get more, you have quite a high number of decoys generate 40-50 or so, whilst one case where you got lower credit than claim, you have 32 decoys generated. I'm not sure if this is based on how long it took for the decoys to be generated, or if there's some other factor as well - but if you've done less work on a particular work-unit, it's only fair that you get less credit, in my opinion [however, that's assuming all decoys are similarly difficult to create, which isn't the case] - but it may be that on your particular machines, some workunits fits nicely in the cache, and others (larger ones) don't - which means that small models will calculate quickly and get high scores, larger models calculate slowly, and get less credit - this would be the natural effect of a "credit per work" system - if you do it more slowly, then you get less credit.

If it makes you any happier, my 24hr range for 70-odd WU's is 292-540 credit, so around 150 points more for some than others... The claimed creidt, naturally, is pretty close to constant - since that's simply based on the amount of time and benchmark result for that machine, and the time doesn't vary very much...

--
Mats

---

cpu id# #of Secs old cred new cred cpu type.
-----------------------------------------------------
263628 570895.46 2164.82 2286.09 Athlon Xp-M
263683 598285.02 2240.28 2164.68 Athlon xp 2600+
283630 541682.72 2024.26 2181.04 Athlon Xp-M
283629 601105.77 2236.36 2351.15 Athlon Xp-M
................... total: 8665.72 8982.96

I totalled the 28-30 results that had been turned in for each of the above machines. Per week, the Atlon XP 2600+ got about 76 points less under the new system that it would have with the old credit system. While the XP-M cpus brought the total to 320 points more than you would have received under the old system for all 4 machines per week. The XP-Ms were getting short changed under the old system; while the Athlon XPs may have been getting a bit more than they should have.

I'm not seeing evidence that the old credit system is/was better for your machines than the new one.
Under the old credit system, Boinc would have given you 1237.96 credits/day. The new credit system is giving you 1283.28 credits/day.

(Neat how this strips out all the nice formatting.. )

---

Steve, have you got any statistics beyond what is currently in your WU list? Looks to me like about 80% of your WUs are within +/-10 points of 105. There's also a 2% variance in your reported crunch time on the WUs, so don't forget to account for that. This is why people often study their credit based on

Perhaps you could express what your expectation of a credit system that doesn't xxxxx. And perhaps that leads to improvements.

---

Add this signature to your EMail:
Running Microsoft's "System Idle Process" will never help cure cancer, AIDS nor Alzheimer's. But running Rosetta@home just might!
https://boinc.bakerlab.org/rosetta/

BoincStats
Some of the charts and lists show the kind of credit being granted for the past few months, rather than just the last week.

I compared it to my single machine.. and during the first week of that graph, he got up to double what he's getting now; while (with the exception of yesterday missing) I've been getting a little more than than the first week.

---

I just love this...

The changers are fighting the facts right in front of them. I pointed to one machine to show that the new credit system xxxxx. Then everyone wants to compare all my machines to even it out. Or supply some formula to tell me what I was getting. I know what I was getting, and I know what I am getting now.

There is no way that a single machine running 8hr wu's should have a +/- 40 point difference, with 2 wu's that were within 60 seconds of each other in run time. This is fact and not fair.

LOOK AND SEE THE FACTS!

https://boinc.bakerlab.org/rosetta/result.php?resultid=39894807
https://boinc.bakerlab.org/rosetta/result.php?resultid=40108533

I don't care how many decoys it ran, it ran for the same time, and the credit is a crap shoot. All this talk of decoys is just that, a decoy. It did the same work within 60 seconds, its a dedicated machine and does nothing else.

And when I first got here from Predictor, I ran the stock client for less than a week, and received just a little over what I was getting at Predictor, and that was just a little over 1500 ppd. Then I saw that you could get 2-3 times the credit running SEE550 so I switched for about 10 days before the new credit system went live, then switched back to the stock client.

And just to clear thing up further, all my machines are Mobile 2600 Baton's. The bios in the one machine sees it as a 2600+ because of the 16 multiplier. This was my second fastest machine in Predictor. (#3)

#1 runs 2635mhz
#2 runs 2610mhz - this machine has been increased by 105 mhz
#3 runs 2597mhz - this is the machine with the results shown
#4 runs 2557mhz - this machine has been increased by 40 mhz
All water cooled XP Mobile Barton 512 L2 cache Core 2600's, running 2-2-2-x memory timings.

And my expectation of a fair credit system is one that rewards the same amount of points for a machine for the same amount of work done.

Example: (was getting over 375 ppd on Predictor)

My machine #1 runs 5 wu's for 8.0 hours and gets 125 credits every time. If it runs 7.9 hrs it gets 123 every time. If it runs 8.1 hrs it gets 127 every time. Its running the same speed 24/7 and doing the same thing, the only difference is it runs 30 long decoys or 50 short decoys. Should make no difference in points awarded. It did its work at the same speed and the only change was the amount of decoys run. Should have no impact on points awarded. If you bump up the speed of the machine by 50mhz you should get an extra 3-5 credits per wu. for the same run times.

Something along these lines would be fair. Not this random crap shoot thats happening now.

---

BETA = Bahhh

Way too many errors, killing both the credit & RAC.

And I still think the (New and Improved) credit system is not ready for prime time...

Given the fact that you can see the old credit system scores, and you're getting 1237 credits/day from the old credit system scores over the last week and the new credit system is giving you MORE than the old credit system would have.... it's obvious that the credit system is not the problem with your having a lower score now than when you moved from Predictor.

There's 86,400 seconds in a day; and 604,800 seconds in a week.
Machine 263628 is missing around 30,000 seconds (an 8 hour WU); while machine 283630 is missing around 60,000 seconds (2ea 8 hour WUs). Since there were no failed WUs, find out what's eating up the cpu time on those two systems, and you'll get closer to the credit/day that you want.

If the bios on the system that can't identify the XP-M correctly - will let you, try running the cpu bus and the memory bus at the same speed. When I got my Athlonxp 2600+, there was talk about mixed cpu/mem bus speeds ruining throughput on some motherboards.

My 24 hour WUs have wild fluctuations as well. When the fluctuations were brought up, we were told that it would balance out with time.

---

I just love this...

The changers are fighting the facts right in front of them. I pointed to one machine to show that the new credit system xxxxx. Then everyone wants to compare all my machines to even it out. Or supply some formula to tell me what I was getting. I know what I was getting, and I know what I am getting now.

There is no way that a single machine running 8hr wu's should have a +/- 40 point difference, with 2 wu's that were within 60 seconds of each other in run time. This is fact and not fair.

LOOK AND SEE THE FACTS!

https://boinc.bakerlab.org/rosetta/result.php?resultid=39894807
https://boinc.bakerlab.org/rosetta/result.php?resultid=40108533

I don't care how many decoys it ran, it ran for the same time, and the credit is a crap shoot. All this talk of decoys is just that, a decoy. It did the same work within 60 seconds, its a dedicated machine and does nothing else.

And when I first got here from Predictor, I ran the stock client for less than a week, and received just a little over what I was getting at Predictor, and that was just a little over 1500 ppd. Then I saw that you could get 2-3 times the credit running SEE550 so I switched for about 10 days before the new credit system went live, then switched back to the stock client.

And just to clear thing up further, all my machines are Mobile 2600 Baton's. The bios in the one machine sees it as a 2600+ because of the 16 multiplier. This was my second fastest machine in Predictor. (#3)

#1 runs 2635mhz
#2 runs 2610mhz - this machine has been increased by 105 mhz
#3 runs 2597mhz - this is the machine with the results shown
#4 runs 2557mhz - this machine has been increased by 40 mhz
All water cooled XP Mobile Barton 512 L2 cache Core 2600's, running 2-2-2-x memory timings.

And my expectation of a fair credit system is one that rewards the same amount of points for a machine for the same amount of work done.

Example: (was getting over 375 ppd on Predictor)

My machine #1 runs 5 wu's for 8.0 hours and gets 125 credits every time. If it runs 7.9 hrs it gets 123 every time. If it runs 8.1 hrs it gets 127 every time. Its running the same speed 24/7 and doing the same thing, the only difference is it runs 30 long decoys or 50 short decoys. Should make no difference in points awarded. It did its work at the same speed and the only change was the amount of decoys run. Should have no impact on points awarded. If you bump up the speed of the machine by 50mhz you should get an extra 3-5 credits per wu. for the same run times.

Something along these lines would be fair. Not this random crap shoot thats happening now.

The Rosetta people are using a credit system that they hope balances out over time. You may get some higher, you may get some lower, but over time they hope when you add them all up and calculate the average they should be pretty much what you should be getting. Unfortunatly we see and look at individual tasks and see the spread (I even reported it over on Ralph before it changed over). I can say that Rosetta@home is not a suitable roject for people that want to measure their CPU speed via the stats, you will not notice the 100MHz increase unless you take the results from before and calculate the average, then compare that to the average from the results after.

Is it fair? That depends on what you call fair. It's fair in the sense we all have to use it and we all share it's problems, It is hoped it is fair in the long run for a copmuter, but it is certainly not fair for individual tasks.

---

Team mauisun.org

Steve all anyone was trying to do was to point out that you picked your machine with the most extreme variance, and the WUs with the most extreme variance and called it a problem.

I challenge your assertion that all of these WUs actually performed the same amount of work during the time that they ran. In fact, the very point that your other machines are more consistent with credit should point you to the idea that even though the WU is "running", it doesn't have the full resources of the machine available all of the time. It is conflicting with something else that is running on that machine.

Example, if two identical machines run identical WUs, and one machine has other tasks kicking in which are taking L2 cache and memory away from the R@H crunching, the WU shows "running" all the time... but you know it is not getting as much actual work done, because it has to endure the resulting L2 cache misses and the page faults.

Another example, two identical machines run identical WUs, but the WUs have different starting seeds (which is generally true). The difference in seed values causes the calculations on the protein to take a different path for some models crunched. That different path may require more floating point calculations, or more memory, or hit some code which conflicts with other areas of the application for L2 cache addresses... if one machine hits this portion of code, it's going to complete less work in a given time, even though it's running full-time, dedicated.

There IS variance in specific models crunched, and the credit system only reflects that in it's running average of credit claims. The credit system is not as granular as to recognize which of your models were found to be uninteresting, and cut off early, and which were crunched through completion. And since it is based upon the BOINC reported credit claim, which is not granular to the decoy level, it may prove difficult to add any further granularity to the system.

Science is not fair. Some people get sick and die, while others live. You are one of the ones that has lived... that's not fair. But we accept the reality of it.

---

Add this signature to your EMail:
Running Microsoft's "System Idle Process" will never help cure cancer, AIDS nor Alzheimer's. But running Rosetta@home just might!
https://boinc.bakerlab.org/rosetta/

yawn

It's ironic that you don't seem to feel all of the variance you observed before the new credit system was in place as being any sort of problem at all. Or a credit system that xxxxx.

---

Add this signature to your EMail:
Running Microsoft's "System Idle Process" will never help cure cancer, AIDS nor Alzheimer's. But running Rosetta@home just might!
https://boinc.bakerlab.org/rosetta/

Balances out over time? What a crock. How about some consistency, and no hap - hazard, throw the darts at a board to assign credit, system.

Under the old credit system with the stock client, I was getting just over 1500 ppd, just the same as Predictor. With the new and improved credit system, random chaos is occurring and my PPD has fallen by over 200, and I have added 145 mhz to the machines since moving to this project, and points should have gone up accordingly.

With the old credit system, there was no +/- 40 point differential on the same 8hr wu's. And the box in question was my second highest producer. That has changed as I have tweaked the speeds up on 2 other boxes, and the credits should have followed. But the fact remains that this box is the 3rd fastest and is getting the lowest scores. Something has changed, and its the credit system, not the rig. The scores on that box changed when the credit system changed.


(quote)(Steve all anyone was trying to do was to point out that you picked your machine with the most extreme variance, and the WUs with the most extreme variance and called it a problem.)

Yep, and it is. That was the point.

(quote)(I challenge your assertion that all of these WUs actually performed the same amount of work during the time that they ran. In fact, the very point that your other machines are more consistent with credit should point you to the idea that even though the WU is "running", it doesn't have the full resources of the machine available all of the time. It is conflicting with something else that is running on that machine.)

Sorry, but all but my main rig are stripped down for crunching only. My main rig only gets used occasionally when the wife's on the laptop. No other apps, spyware, viruses, etc. And there are no conflicts. All CPU temps are under 47c under load. This is my point, somethings wrong and it is not on my end. This box was very consistent on Predictor, meaning something is not right here. And the facts back up my original claim of +/- 40.

All boxes are running N-Force chipsets and fsb and memory are 1 to 1. I know how to overclock, and that tight 2-2-2-x timings are key to memory bandwidth.

(quote)(yawn) = BAIT. Nice. And I see that you are one of the people with more posts than RAC. As a matter of fact all the people that are defending this system have more posts than RAC. And the 2 people that agree have more RAC than posts. Interesting...is it not... And with the baiting, my response will be no response.


Since this is now an attack on my views, and proven facts, and seem to have been called a liar, I will not be posting replys to this thread.

I could have helped to solve this, but I will not be drawn into a baiting flame war, or be called a liar. I thought that this forum had reached a level of maturity to discuss this matter. I see I was wrong.

For me this matter is now dead, along with this project. I will only propose a tiny solution. You cannot have more posts than RAC. This would force people into choosing when and what they post more carefully. = yawn

I will continue to crunch this project for its contribution to the science of disease cures. And my team. When Predictor comes back online I will go back there for fair credit. I can see it is not going to happen here.

I'm done.....

---

BETA = Bahhh

Way too many errors, killing both the credit & RAC.

And I still think the (New and Improved) credit system is not ready for prime time...

Biggest problem we have here is you are doing this for the wrong reason. You want the prestige of a high RAC and high score. Well I only have 1 machine a rac of about 130ish i do this crunching for 1 reason alone.Maybe 1 day a cure for something will come out of the work. So get off the megalo mania of Points and crunch for the real reason to help others. Enough said. Oh i have looked at some off my old scores and i would get varied returns and the machine was running the same all the time. So as the others have pointed out some variables inthe Wus do happen live with it


Charlie

---

Balances out over time? What a crock. How about some consistency, and no hap - hazard, throw the darts at a board to assign credit, system.

Under the old credit system with the stock client, I was getting just over 1500 ppd, just the same as Predictor. With the new and improved credit system, random chaos is occurring and my PPD has fallen by over 200, and I have added 145 mhz to the machines since moving to this project, and points should have gone up accordingly.

With the old credit system, there was no +/- 40 point differential on the same 8hr wu's. And the box in question was my second highest producer. That has changed as I have tweaked the speeds up on 2 other boxes, and the credits should have followed. But the fact remains that this box is the 3rd fastest and is getting the lowest scores. Something has changed, and its the credit system, not the rig. The scores on that box changed when the credit system changed.


(quote)(Steve all anyone was trying to do was to point out that you picked your machine with the most extreme variance, and the WUs with the most extreme variance and called it a problem.)

Yep, and it is. That was the point.

(quote)(I challenge your assertion that all of these WUs actually performed the same amount of work during the time that they ran. In fact, the very point that your other machines are more consistent with credit should point you to the idea that even though the WU is "running", it doesn't have the full resources of the machine available all of the time. It is conflicting with something else that is running on that machine.)

Sorry, but all but my main rig are stripped down for crunching only. My main rig only gets used occasionally when the wife's on the laptop. No other apps, spyware, viruses, etc. And there are no conflicts. All CPU temps are under 47c under load. This is my point, somethings wrong and it is not on my end. This box was very consistent on Predictor, meaning something is not right here. And the facts back up my original claim of +/- 40.

All boxes are running N-Force chipsets and fsb and memory are 1 to 1. I know how to overclock, and that tight 2-2-2-x timings are key to memory bandwidth.

(quote)(yawn) = BAIT. Nice. And I see that you are one of the people with more posts than RAC. As a matter of fact all the people that are defending this system have more posts than RAC. And the 2 people that agree have more RAC than posts. Interesting...is it not... And with the baiting, my response will be no response.


Since this is now an attack on my views, and proven facts, and seem to have been called a liar, I will not be posting replys to this thread.

I could have helped to solve this, but I will not be drawn into a baiting flame war, or be called a liar. I thought that this forum had reached a level of maturity to discuss this matter. I see I was wrong.

For me this matter is now dead, along with this project. I will only propose a tiny solution. You cannot have more posts than RAC. This would force people into choosing when and what they post more carefully. = yawn

I will continue to crunch this project for its contribution to the science of disease cures. And my team. When Predictor comes back online I will go back there for fair credit. I can see it is not going to happen here.

I'm done.....

Hello, you should check your history of RAC as well, I used to have a RAC near 1500, but the cost of electricity and the increased demands of Rosetta@home (7 day deadline) have caused it to drop, that coupled with running and testing some other projects due to the alterations and decitions by moderators and the credit system I have increased testing other project for fun.

I have also brought far more people contributing far higher RAC and lower to this project and teams than you know.


I said it as it is, we have to live with it. If your computer is gaining slower credit than your others now it because of something like a slower memory bus? a smaller cache size [not applicable in your case], maybe even one of the other memory timings, though likely just pot luck :-(. That is the way it works now and we all have to live with it for the time being if you run Rosetta@Home.
Thank youself you haven't got a PPC/Mac, they have dropped too at least half their standard RAC, but that is due to how poorly they do Rosetta@home compared to what the benchmark scores.

If you want credit, go to Einstien they give out the most, or put and benchmark altering (or alter the cleint_state file yourself and run one of the other smaller projects ;-)
You may want to look at docking@home that has just started a larger beta testing or WCG or maybe SIMAP.

Is the current credit system flawed, like I said yes. Early returned task in a task group can gain higher or lower credit as they claim far nearer their claimed (I had one I noticed). Individual tasks have fluctuations in credits, this is due to the varying lengths the decoys that the sytem is based on.
Like I said it flawed on the small scale but should, the project hopes, even out in the large scale. Again it is not a good project if you want to see which computer can perform a Whetstone/Drystone benchmark the fastest.


Suggest a way to make it better system ?
- They cannot use redundency, they have no scientific reason to use it (so they say)
- They cannot use the boinc banchmark (you get what you claim and it is so easy to just put whatever number you want for the benchmark into the client_state file)

- We have suggested internal benchmarks in the science appliaction based on the science itself, including a internal timing process as well.
- some others...

Your a frog, the frogs have ran DC projects I have been active in they are a good bunch and normally willing to give solutions to problem, (well at least have some fun in the converstaions ;-)


River,
that is an inappropriate comment and something out newer mods would normaly remove as flaim bait.

---

Team mauisun.org

Stevea:
The large variance of some of my 86400 second WUs inspired me to post suggestions on minimizing the variance. It bugged me too, at first.

I was interested in finding out why you had a drop in daily credit. As far as I was aware, Macs, Durons, and Celerons are the only ones that are producing at a rate lower than their benchmarks (claimed credit) would suggest. You are now getting 1236/1500= 82.4% of what you claim you were getting from the old system. The claimed credit IS the old credit system. I was getting around 260 credits for 86400 second WUs before the new credit system.. I have a record of 7 of them from 8/26, and I'm still getting around 260 credits for 86400 second WUs now. There has not been a 20% decrease in the claimed credit on my machine as it's still the same.

BoincStats claims you've got 5 systems associated with your account; we're only seeing 4 now. Is it possible that you were running a 5th system when you first moved over? Another possibility for higher than actual RAC is dumping a large amount of WUs in a short period of time (saving up 2 days worth, and then dumping them at once..) so that it takes awhile for your RAC to decline to what you're actually producing per day.

Machine id#263683 may be your second fastest benchmarked machine - but if it was producing the same amount of work as the other 3 machines, it would be getting 1.06 times the claimed credit that the other machines are getting. Since it's getting 0.966 times the claimed credit, I'd recommend getting another of the motherboards used by the other 3 machines, flash the same bios, and get this system running at 1.06 times the claimed credit.

Any luck tracking down the reason for the 3 missing 8 hour WUs from the two machines? i.e. is Rosetta being shut off while something else is running? Tracking down the missing 24 hours and setting machine id#263683 up identical to the other 3 will result in a nearly 100ppd increase.

---

Stevea:
The large variance of some of my 86400 second WUs inspired me to post suggestions on minimizing the variance. It bugged me too, at first.

I was interested in finding out why you had a drop in daily credit. As far as I was aware, Macs, Durons, and Celerons are the only ones that are producing at a rate lower than their benchmarks (claimed credit) would suggest. You are now getting 1236/1500= 82.4% of what you claim you were getting from the old system. The claimed credit IS the old credit system. I was getting around 260 credits for 86400 second WUs before the new credit system.. I have a record of 7 of them from 8/26, and I'm still getting around 260 credits for 86400 second WUs now. There has not been a 20% decrease in the claimed credit on my machine as it's still the same.

BoincStats claims you've got 5 systems associated with your account; we're only seeing 4 now. Is it possible that you were running a 5th system when you first moved over? Another possibility for higher than actual RAC is dumping a large amount of WUs in a short period of time (saving up 2 days worth, and then dumping them at once..) so that it takes awhile for your RAC to decline to what you're actually producing per day.

Machine id#263683 may be your second fastest benchmarked machine - but if it was producing the same amount of work as the other 3 machines, it would be getting 1.06 times the claimed credit that the other machines are getting. Since it's getting 0.966 times the claimed credit, I'd recommend getting another of the motherboards used by the other 3 machines, flash the same bios, and get this system running at 1.06 times the claimed credit.

Any luck tracking down the reason for the 3 missing 8 hour WUs from the two machines? i.e. is Rosetta being shut off while something else is running? Tracking down the missing 24 hours and setting machine id#263683 up identical to the other 3 will result in a nearly 100ppd increase.

Benny, there's 6 associated, click on show all ;-)

---

Team mauisun.org

Steve, challenging your assumptions and stating very specifically where I feel your assumptions are flawed certainly should not be constrewed as an attack. You clearly know a lot about hardware and overclocking. But there's more to it then that. The science done in each WU differs. The random number used for a given model tries a different path. It seems your knowledge of the WUs and these variances is lacking.


+/- 40 is nothing compared to the 3x difference between someone running a non-standard client that claims higher credit, and someone with an identical machine running the standard client. The old credit system would give someone with an identical machine triple the RAC you are getting with your standard clients Steve, even though your machine produces the same output as the other person. So, please agree that the new credit system is flawed, has room for improvement, but is far more fair then the old system. And therefore the credit system is showing improvement over time.

---

Add this signature to your EMail:
Running Microsoft's "System Idle Process" will never help cure cancer, AIDS nor Alzheimer's. But running Rosetta@home just might!
https://boinc.bakerlab.org/rosetta/

River,
that is an inappropriate comment

in which case I apologise.

I do want to say that I felt/feel frustrated by the way the pro- and anti- change people seem (seem to me) continually to argue past each other.

I feel frustrated by the way some (not all) of the posts in the second half of this thread seem (seem to me) to use new words to say the same things as the same people said in the first half of this thread.

Stevea says one of his machines shows a high variance in credit per hour.

Others say the law of averages should make it come right in the medium to long term.

Both sets of facts are totally correct, the gap is not in the facts but in how each of you feels about the facts.

Nobody will convince Stevea that the new system doesn't xxxxx if they try to do so starting from their own priorities instead of from Stevea's.

Stevea will not convince the supporters of the new system that it does xxxxx if he continues to assume they all share his priorities once they have already made it clear their priorities are different.

What I personally would be interested in (not having a view either way) would be if the debate could move along from "proving" the new system or the old system does or did this that and the other, which we all understand. If the debate could move along to looking at the underlying priorities of what we want a credit system to do, and how we prioritise among the things we want it to do.

If we agree at that level we can then work out objectively which system best fits the agreed priorities, but only then.

If we find we need to agree to differ on the priorities then there is no point at all trying to raise the details in debate. The same details applied to differing priorites will inevitably "prove" both sides right on their own terms.

R~~

It is time to get the garments ready:





Again!!!!! YAwn!!!!!

---

This and no other is the root from which a Tyrant springs; when he first appears he is a protector.”
Plato