Changes

:::ASchlafly's statement, "''In fact, while the paper states the generation periods for the First (replay) Experiment, it does not disclose the generation periods for the Second and Third Experiments''" demonstrates a surprising lack of comprehension of the procedures described for the second and third replay experiments. There were no generation periods for the second and third experiments, because the bacteria in those experiments were not serially cultured. They were plated once onto MC agar (which contains no glucose) and allowed to sit on those same plates for 59 days (replay 2) or 49 days (replay 3).  

:::ASchlafly's statement, "''In fact, while the paper states the generation periods for the First (replay) Experiment, it does not disclose the generation periods for the Second and Third Experiments''" demonstrates a surprising lack of comprehension of the procedures described for the second and third replay experiments. There were no generation periods for the second and third experiments, because the bacteria in those experiments were not serially cultured. They were plated once onto MC agar (which contains no glucose) and allowed to sit on those same plates for 59 days (replay 2) or 49 days (replay 3).  

::: If you continue your insulting tone, then you'll join others who have been blocked.  If you're claiming that the Second and Third replay Experiments did not have new generations during their 59 and 49 day periods, then please clearly say so.  Note that Lenski does not observe when the Cit+ variants were allegedly observed during those periods.--[[User:Aschlafly|Aschlafly]] 23:07, 24 July 2008 (EDT)

:::: If you continue your insulting tone, then you'll join others who have been blocked.  If you're claiming that the Second and Third replay Experiments did not have new generations during their 59 and 49 day periods, then please clearly say so.  Note that Lenski does not observe when the Cit+ variants were allegedly observed during those periods.--[[User:Aschlafly|Aschlafly]] 23:07, 24 July 2008 (EDT)

 

::::The second and third replay experiments did not have 'generations' in the way that the LTEE and the first replay experiment did. Some of the Cit- cells ''may'' have divided once or twice before they ran out of stored glucose, but for the vast majority of the 59 or 49 days they were sitting dormant on the plates until they became Cit+, at which point they started to form colonies. Lenski ''does'' state that Cit+ mutant colonies were noted between 8 and 28 days after plating, and states that control plates that started out with a mix of Cit+ and Cit- cells showed Cit+ colonies after two to three days. Furthermore, the paper points out that when the new Cit+ cells are replated on MC agar, they also form visible colonies after 2 days - so they are not inherently slow-growing. --[[User:Brossa|Brossa]] 10:22, 25 July 2008 (EDT)

:::Similarly, his statement "''The same reason that the paper admits an inability to "exclude an earlier origin" (p. 7901) for the Cit+ variants also results in an inability to exclude Cit+ variants from the samples taken after generation 31,000''" is also wrong. It is impossible to exclude an earlier origin ''in a population of millions of cells'' because it is impractical to plate all of the cells for each generation out so thinly that each and every cell gives rise to a unique colony which can then be tested for the Cit+ trait. For each generation you would have to run thousands of plates to determine ''exactly'' when the first Cit+ cell arose. Conversely, if you wish to sample only a few hundred or thousand cells, rather than multiple millions, it is straightforward to plate them out at such a great dilution that every cell in the sample gives rise to a unique colony, separated in space from all the others, which can then be tested individually for the traits you wish to identify. Since each colony arises from a single cell in the sample, after generation 31,000 you will have colonies that are entirely Cit+ and those that are entirely Cit-. His claim that "''there is no scientific basis for including these Cit+ populations in this study, and it only serves to distort the results''" is also wrong - there is a basis to include those populations - or to be more specific, the Cit- subset of those populations. At the time the replay experiments were designed, before there was any evidence for a potentiating mutation at 20,000 generations, it was impossible to know which generations were more likely to give rise to the Cit+ trait. Suppose that a potentiating mutation arose at the 25,000th generation and expanded through the population or simply persisted at a low level through the 31,000th generation, ''by which time it was outcompeted by some other mutation, or was lost through drift''. It could have been the case that the Cit+ mutation could ONLY have arisen between, say, the 25,000th and 31,500th generations, with no 'potentiated-but-not-Cit+' cells lasting past the 31,500th generation. As it turns out, it appears that the potentiated cells continue to persist in the Cit- population through the 32,500th generation. If you are trying to determine whether the Cit+ trait is contingent or simply random, it absolutely makes sense to include Cit- cells of later generations.

:::Similarly, his statement "''The same reason that the paper admits an inability to "exclude an earlier origin" (p. 7901) for the Cit+ variants also results in an inability to exclude Cit+ variants from the samples taken after generation 31,000''" is also wrong. It is impossible to exclude an earlier origin ''in a population of millions of cells'' because it is impractical to plate all of the cells for each generation out so thinly that each and every cell gives rise to a unique colony which can then be tested for the Cit+ trait. For each generation you would have to run thousands of plates to determine ''exactly'' when the first Cit+ cell arose. Conversely, if you wish to sample only a few hundred or thousand cells, rather than multiple millions, it is straightforward to plate them out at such a great dilution that every cell in the sample gives rise to a unique colony, separated in space from all the others, which can then be tested individually for the traits you wish to identify. Since each colony arises from a single cell in the sample, after generation 31,000 you will have colonies that are entirely Cit+ and those that are entirely Cit-. His claim that "''there is no scientific basis for including these Cit+ populations in this study, and it only serves to distort the results''" is also wrong - there is a basis to include those populations - or to be more specific, the Cit- subset of those populations. At the time the replay experiments were designed, before there was any evidence for a potentiating mutation at 20,000 generations, it was impossible to know which generations were more likely to give rise to the Cit+ trait. Suppose that a potentiating mutation arose at the 25,000th generation and expanded through the population or simply persisted at a low level through the 31,000th generation, ''by which time it was outcompeted by some other mutation, or was lost through drift''. It could have been the case that the Cit+ mutation could ONLY have arisen between, say, the 25,000th and 31,500th generations, with no 'potentiated-but-not-Cit+' cells lasting past the 31,500th generation. As it turns out, it appears that the potentiated cells continue to persist in the Cit- population through the 32,500th generation. If you are trying to determine whether the Cit+ trait is contingent or simply random, it absolutely makes sense to include Cit- cells of later generations.

:::: The alleged potentiating mutation must have occurred prior to the 31,500th generation, and it makes no sense to test subsequent generations for the potentiating mutation.  Rather, including later generations only distorts the statistical analysis.--[[User:Aschlafly|Aschlafly]] 23:07, 24 July 2008 (EDT)

:::: The alleged potentiating mutation must have occurred prior to the 31,500th generation, and it makes no sense to test subsequent generations for the potentiating mutation.  Rather, including later generations only distorts the statistical analysis.--[[User:Aschlafly|Aschlafly]] 23:07, 24 July 2008 (EDT)

:::His statement "''The paper incorrectly combined the Third Experiment with the other two based on outcome rather than sample size, thereby yielding a false claim of overall statistical significance''" is meaningless. If ASchlafly objects to the statistical method used, he should recalculate the results using his preferred test and then present both his results and a defense of the alternate statistical method. If you send a letter to PNAS that includes such obvious errors of fact and unsupported conclusions, it will only embarrass ASchlafly and Conservapedia further.--[[User:Brossa|Brossa]] 16:20, 24 July 2008 (EDT)

:::His statement "''The paper incorrectly combined the Third Experiment with the other two based on outcome rather than sample size, thereby yielding a false claim of overall statistical significance''" is meaningless. If ASchlafly objects to the statistical method used, he should recalculate the results using his preferred test and then present both his results and a defense of the alternate statistical method. If you send a letter to PNAS that includes such obvious errors of fact and unsupported conclusions, it will only embarrass ASchlafly and Conservapedia further.--[[User:Brossa|Brossa]] 16:20, 24 July 2008 (EDT)

:::: You resort to insults when you lack substance.  One need not propose an alternative, or a solution, in order to identify a flaw.  One may announce that a bridge is defective and should be closed prior to figuring out how it can be fixed, if indeed it can be fixed.--[[User:Aschlafly|Aschlafly]] 23:07, 24 July 2008 (EDT)

:::: You resort to insults when you lack substance.  One need not propose an alternative, or a solution, in order to identify a flaw.  One may announce that a bridge is defective and should be closed prior to figuring out how it can be fixed, if indeed it can be fixed.--[[User:Aschlafly|Aschlafly]] 23:07, 24 July 2008 (EDT)

::::: I agree that you don't really need to use an alternative statistical test ''per se'' to identify an alleged flaw. However, you do need to meet statistics with statistics and use the method that was used in the paper (the Monte Carlo algorithms, preferably with the same software) to show quantitatively where the error was made and what you think the result should have been. The reason I suggest discussing your result is that in most correspondence of this nature in scientific journals, the discussion is usually something like "Well, the authors calculated this result, but we recalculated and got something different. This is why we think that happened." So while I don't think you need to propose an alternative technique, you should discuss what the result should have been (using the data in Table 1) as that is the standard approach to such concerns. To borrow from your analogy, you don't need to figure out how the bridge can be fixed before announcing it is defective, as you said, but announcing defectiveness isn't enough to close it- you must ''show'' that it is. Hope that helps.[[User:Kallium|Kallium]] 09:52, 25 July 2008 (EDT)

::::: I agree that you don't really need to use an alternative statistical test ''per se'' to identify an alleged flaw. However, you do need to meet statistics with statistics and use the method that was used in the paper (the Monte Carlo algorithms, preferably with the same software) to show quantitatively where the error was made and what you think the result should have been. The reason I suggest discussing your result is that in most correspondence of this nature in scientific journals, the discussion is usually something like "Well, the authors calculated this result, but we recalculated and got something different. This is why we think that happened." So while I don't think you need to propose an alternative technique, you should discuss what the result should have been (using the data in Table 1) as that is the standard approach to such concerns. To borrow from your analogy, you don't need to figure out how the bridge can be fixed before announcing it is defective, as you said, but announcing defectiveness isn't enough to close it- you must ''show'' that it is. Hope that helps.[[User:Kallium|Kallium]] 09:52, 25 July 2008 (EDT)