Please refer to this WU:
http://setiweb.ssl.berkeley.edu/workunit.php?wuid=11989121

4 of us appear to have crunched with success, I'm not sure why the 5th WU was uploaded on 12 April as 4 successful results were in on the 10th.

The 6th result ID thingie is unsent (as at the timing of this post).

Any ideas on what is happening with closing off this WU? There don't appear to be any errors associated with the ul'ed results.

cheers

> 4 of us appear to have crunched with success, I'm not sure why the 5th WU was
> uploaded on 12 April as 4 successful results were in on the 10th.

Seems that validator thinks not all returned results are withing validation margins. THerefore new results are needed so that consensus (3 results that agree enough) can be formed.

All the results have this state:

Checked, but no consensus yet

---

Metod ...

The jury is still out, so to speak...

---

> The jury is still out, so to speak...

As long as it's not ruled a mistrial and restarted, I shall keep silent on this matter. :)

> 4 of us appear to have crunched with success, I'm not sure why the 5th WU was
> uploaded on 12 April as 4 successful results were in on the 10th.
>
> The 6th result ID thingie is unsent (as at the timing of this post).

I Had a similar one recently, http://setiweb.ssl.berkeley.edu/workunit.php?wuid=12018014

It was not validated until 5 results were in and then all five were given credit, strange but true. When I asked in a similar thread I was led to believe that in the first four results that no three results were close enough to be validated and that a fifth, and maybe more, would be sent until a least three were within limits.
How four results cannot be close enough and then all five are given credit escapes me.

>
> It was not validated until 5 results were in and then all five were given
> credit, strange but true. When I asked in a similar thread I was led to
> believe that in the first four results that no three results were close enough
> to be validated and that a fifth, and maybe more, would be sent until a least
> three were within limits.
> How four results cannot be close enough and then all five are given credit
> escapes me.
>

Still not saying anything about the seti-validator, 4 not validated but 5 all validated is easily explained by example-validator needing 3 results with error-limit +-1.
You've got 4 results at 28, 28.1, 29.9 and 30. For these, there is no 3 results inside the error-limit, so fails validation.
If 5th result is 29, all 5 results is inside error-limit so all 5 validated.

> How four results cannot be close enough and then all five are given credit
> escapes me.

I have an example that shows the "basket" and it is possible that as the results came in they could not compare, but later "filled-the-gaps" and allowed all 5 to be in the basket and thusly granted crredit.

Again, I know, somewhat conunter intuitive ...

---

> Still not saying anything about the seti-validator, 4 not validated but 5 all
> validated is easily explained by example-validator needing 3 results with
> error-limit +-1.
> You've got 4 results at 28, 28.1, 29.9 and 30. For these, there is no 3
> results inside the error-limit, so fails validation.
> If 5th result is 29, all 5 results is inside error-limit so all 5 validated.
>
When I asked in the previous thread about no validation on four results, I asked if this was because there were two pairs close together, one pair and two separate results or four separate results. The very simple answer I was given was '4'. I therefore assumed that it was not like the example given above which I regard as two pairs, which can be joined by subsequent result that is in the middle. I had assumed that with your answer being '4' then at least one result would be rejected, as no result was rejected and all five were given credit, I remained confused, and to a certain extent still am.

> When I asked in the previous thread about no validation on four results, I
> asked if this was because there were two pairs close together, one pair and
> two separate results or four separate results. The very simple answer I was
> given was '4'. I therefore assumed that it was not like the example given
> above which I regard as two pairs, which can be joined by subsequent result
> that is in the middle. I had assumed that with your answer being '4' then at
> least one result would be rejected, as no result was rejected and all five
> were given credit, I remained confused, and to a certain extent still am.

When we talk of a result we are talking a multi-dimensional entity. We simplify our examples by ignoring the actual complexity of the answers.

In gross terms what I am saying is that when we are done processing a SETI@Home WOrk Unit, we return a file which has a list of our findings, how many Gaussians, how many spikes, how many triplets, and data about each and every one of these.

Yet, we talk of the result as though there was not this complexity.

And when the Result you send back is compared to the Result I send back and to another Result sent back by some one else ... well ... it is not quite so simple ...

But, the example above does fit the "facts" reasonably well ... if we suppose that we had two sets of separated "Results" (most easily pictured as I did with spikes in my examples) the "basket" would not necessarily fit over the two pairs in a "natural" way. Yet, when the 5th example came in, it might easily have fallen into the "middle" and formed a "bridge" that then allowed all of the Results to be found within the "basket".

Again, we are using simplified models to try to explain how this multi-dimensional comparison might be accomplished. In addition, I would add that the "error bounds" of the "basket" may be dependent on the data itself. Meaning what (you might ask)? Meaning, quite simply, that with 5 Results the "basket" is larger ... allowing additional tolerance for inclusion of the 5 Results returned.

Note, I am not saying that this is how it is done, only that this is illustrative of how it may be done.

---

I too have got a pending result from early April, three results are already in and the fourth due earlier today. Some of my credits have been done with only 3 results so why not this one?

---

> I too have got a pending result from early April, three results are already in
> and the fourth due earlier today. Some of my credits have been done with only
> 3 results so why not this one?

I got one that is pending since July 2004 ... with one of the Results tagged as "Didn't need" ...

WIthout knowing which work unit you are asking about it is hard to tell you why. But the usual answer is that the Results were not the same, one or more are Invalid, etc. ...

This one is interesting in that it has pending for 3, and it too seems abandoned.

---

>
> I got one that is pending
> since July 2004 ... with one of the Results tagged as "Didn't need" ...

There's still some old dead wu listed in the database, either due to bugs already fixed or crashes, so till get the time to remove them they'll still show up.

>
> This one is
> interesting in that it has pending for 3, and it too seems abandoned.
>

It is not abandoned, if you takes a little closer look you see a new "result" was generated immediately after 4th result past its deadline. Due to there being roughly 500k "results" at all times waiting to be sent out, it takes currently 1-2 days from a "result" is generated till it is sent out.

The Validator doesn't trigger generating of 1 more "result" on "no consensus yet", except if there's no more "active" results, therefore the system was waiting on the 4th result before a new one was generated.

> >
> > I got one that is <a> href="http://setiweb.ssl.berkeley.edu/workunit.php?wuid=225087">pending[/url]
> > since July 2004 ... with one of the Results tagged as "Didn't need" ...
>
> There's still some old dead wu listed in the database, either due to bugs
> already fixed or crashes, so till get the time to remove them they'll still
> show up.

Tell that to the people still waiting for credit to be granted on this workunit. State is still initial. It is almost a year since it's been sent out, it has 3 perfectly good claims all in the ballpark.

---

>
> Tell that to the people still waiting for credit to be granted on this
> workunit. State is still initial. It is almost a year since it's been sent
> out, it has 3 perfectly good claims all in the ballpark.
>

Already errored-out wu will never give any credit.

As for not getting credit in the 1st place for this wu, it's very likely due to "no consensus yet", since it was re-issued by a fix-script 02.12.

BTW, the validator doesn't look on the claimed credits at all, till after wu has been validated.