 We have pivots talking about pivot subtitles and pivot subtitling templates that are sometimes used when you don't have the subtitles who can go directly or you don't or it's just easier not to look for them and just translate through English. What are other and perhaps some specific challenges in your opinion in pivot language translation when it comes to achieving consistency? I'm working on an article at the moment that analyzes an Egyptian translation into Polish through English on the one hand, that's the subs, but the voiceover was done directly from Egyptian to Polish. Both of these are available on Netflix and both offer slightly different experience to the viewers because pivot was used in one and the other one is a direct translation. That's a very interesting case study. Perhaps we can go into more detail on that. So this is I understand this is a series in Arabic in the Egyptian variety of Arabic. And so you analyze two modes of translation, subtitles and voiceover or the lack of dubbing. And one of them you're saying was done directly. So the translator knew Arabic. And the other was done with the use of a pivot subtitling templates. So the subtitle only you English. I assume. And what was the result of that? Unsurprisingly, the direct translation from Arabic to Polish is much more faithful, much more accurate than the pivot. But the problem is that the English pivot is not a good translation. I counted errors, for instance, and inaccuracies and so on. And there's the text is divided into segments. So I use this as a unit of measurement. And I counted segments which contain some errors inaccuracies and so on some shifts. And it turned out that the direct Polish translation contains about 1% of segments with errors. The Polish subtitles, which were created from English subs contain 4. something per sense of segments with errors. And the English pivot contains 3. something. So you can see that there is a difference of if you disregard the errors that the translator that the Polish subtitle copied from the pivot, the number of errors is more or less 1% in both. So the problem is that the templates or the pivot that the Polish subtitle worked with is far from perfect. The subtitle was, well, was powerless. He was unable to spot the errors in the pivot. So he copied them, he transferred them into his own text. But his own share of problems is very similar to the share of problems that the direct translator had. We always, we always make mistakes, you know, no translation is perfect. I think that a 99% accuracy I think is very high. There's this NTR model by Pablo Romero Fresco. It's, well, it's for life subtitling, but it's translation. And you might be more knowledgeable than I am, in that case. But the threshold of acceptability is 98%. And I think that's, that's a very reasonable, there's no, at least as far as I know, there's no acceptability threshold in regular interlingual interlingual subtitling. There is a different model, the FAR model by Jan Pedersen. But he doesn't suggest any, any threshold. But I think 98% is, this is, is really good. So if you add, you know, one share of 1%, if you know, 1% of subs contain some, some problem, I think that the quality is very, very high. And here you had, you know, 1% versus almost 5%. But 4% more or less were in the, in the English version.