22 Burst results for "Adword"
The Ben Shapiro Show
"adword" Discussed on The Ben Shapiro Show
"The worst financial market in years, hiring freeze everywhere, Q four is looking great. So first of all, I do love the reaction from people on the left Robert Reich did the same thing. Another billionaire is going to, okay, I would like for you to name a giant tech company that is not owned by a multi multi millionaire. Really, name a big tech company that is not owned by somebody who's with hundreds of millions of dollars. They do not exist. There's no Twitter app that is owned by the homeless guy at the end of the street on grand in Los Angeles. That's not how any of this works. Twitter employees heard news of the latest development in Musk's potential purchase while they were in a meeting, discussing the company's goals in 2023, according to Bloomberg. Roman Chad hurry, the Twitter director of machine learning ethics transparency and accountability tweeted quote living the plot of Succession is effing exhausting. Well, I think this person will not have a job for a particularly long. This is just my quick prediction. Again, director of machine learning ethics transparency and accountability means presumably the director of censorship over at Twitter. And of course, CNN's Daniel Sullivan then tweeted this Twitter employee sums up what I'm hearing from folks inside the company today. Yes, because as it turns out, Elon Musk has different ideas about censorship than you do. And this is why you are so upset. EJ Samson, who works in global business marketing to encourage every Twitter employee to go outside and take a walk. Twitter's head of global development on the Weinstein simply indicated what a roller coaster it has all been. Well, you know what? The good news is that Musk is actually creating a self cleansing mechanism here. It's what or by buying, presumably all these people either out themselves and be fired or quit. And it'll be great, because Twitter needs to be completely restaffed. There's a deep state over at Twitter. In the same way there was a deep state in the federal government, meaning an entrenched bureaucracy that has ideas of its own. And all of those people can simply go away. That would be wonderful. Kevin ruse, who's a longtime critic of open speech on the Internet. He's making some predictions over at The New York Times and what exactly is going to happen now that Elon Musk takes over. Presumably he's going to clean house starting with firing, Twitter's chief executive pira agarwal, according to roose juicy set of text messages between Musk and his friends and business associates emerged last week as part of the legal battle in them. Musk made clear he was unhappy current leadership in particular with Bragg agarwal, the chief executive who took over last year from Jack Dorsey, the text showed that AdWord had initially sought to work constructively with Musk, but the two men eventually clashed, admiral one point told must be a text message his habit of tweeting things like his Twitter dying was not helping me make a Twitter better. What did you get done this week? Musk shot back. This is a waste of time. Yeah, man. Well, good, good, because all these people need to go. Ruth says that the employees will revolt. Well, oh, oh no, so really, go
Data Skeptic
"adword" Discussed on Data Skeptic
"We <Speech_Male> <SpeakerChange> talked about, but <Speech_Male> if <Speech_Male> an ad wins the auction, <Speech_Male> they'll <Speech_Male> also pay the reduced <Speech_Male> bit. <Speech_Male> So when we were <Speech_Male> talking about the algorithm, <Speech_Male> the amount <Speech_Male> that the ad <Speech_Male> actually pays to the <Speech_Male> platform, that does not <Speech_Male> change. That is equal to <Speech_Male> the bit. But in <Speech_Male> this bid <Speech_Male> sharing bit scaling <Speech_Male> idea, both the bid <Speech_Male> is reduced but then the <Speech_Male> payment is also <Speech_Male> reduced and the same <Speech_Male> as the reduced width. So <Speech_Male> these are two <Speech_Male> high level ideas that <Speech_Male> I believe are <Speech_Male> implemented to <Speech_Male> better control <Speech_Male> in this budget <Speech_Male> constrained setting, <Speech_Male> sort of sequential <Speech_Male> auction setting, <Speech_Male> the utilization <Speech_Male> of <Speech_Male> ad budgets. <Speech_Male> And the arose really, <Speech_Male> as far as I understand <Speech_Male> it, or ever inspired <Speech_Male> by <Speech_Male> this classic paper <Speech_Male> on AdWords. <SpeakerChange> <Speech_Male> I'm also <Speech_Male> curious as few would <Speech_Male> be willing to speculate <Speech_Male> a bit about the future <Speech_Male> and where things are headed. <Speech_Male> <Speech_Male> Do insights <Speech_Male> like those coming <Speech_Male> out of your research. <Speech_Male> Are these just going to <Speech_Male> make it more efficient <Speech_Male> for the platforms <Speech_Male> to deliver <Speech_Male> ads or <Speech_Male> might it be that <Speech_Male> even the consumers <Speech_Male> appreciation <Speech_Male> or experience of <Speech_Male> ads becomes <Speech_Male> enhanced in <Speech_Male> some way as <SpeakerChange> well? <Speech_Male> I think these <Speech_Male> ideas algorithmic ideas <Speech_Male> can be used to <Speech_Male> do either. So <Speech_Male> if a business manager <Speech_Male> decides that <Speech_Male> my objective <Speech_Male> is to maximize revenue, <Speech_Male> they can use such an <Speech_Male> algorithm to maximize <Speech_Male> or improve their revenue. <Speech_Male> They <Speech_Male> decide that their <Speech_Male> objective <Speech_Male> is, let's say, just <Speech_Male> to maximize <Speech_Male> click or click through rates <Speech_Male> or expect number <Speech_Male> of clicks, they <Speech_Male> can certainly use these algorithms <Speech_Male> to do better on that. <Speech_Male> So <Speech_Male> these algorithms <Speech_Male> are in that sense <Speech_Male> moldable. <Speech_Male> They can help <Speech_Male> you maximize your <Speech_Male> objective and typically <Speech_Male> my <Speech_Male> understanding is that <Speech_Male> platforms <Speech_Male> don't just want to maximize <Speech_Male> one or the other. They want to <Speech_Male> keep their <Speech_Male> cash flowing, but at the <Speech_Male> same time, also want <Speech_Male> to give <Speech_Male> regard to the long-term <Speech_Male> and maximize <Speech_Male> user satisfaction. <Speech_Male> I think <Speech_Male> it's always going to be <Speech_Male> the synergy between <Speech_Male> both of those things <Speech_Male> that you want to make <Speech_Male> sure that <Speech_Male> you are making <Speech_Male> revenue. But at the same time, <Speech_Male> not sacrificing <Speech_Music_Male> on customers <Speech_Male> and really making them <Speech_Music_Male> lawyer long-term <Speech_Male> customers is <Speech_Male> what's <Speech_Male> going to <Speech_Male> be the future, meaning <Speech_Male> try to do the best you can <Speech_Male> on both those fronts. <Speech_Male> And these kinds of <Speech_Male> algorithms can <Speech_Male> tune <Speech_Male> <Speech_Male> to <Speech_Male> your objective <Speech_Male> in at <SpeakerChange> least <Speech_Male> to some extent. <Speech_Male> Over can listeners <Speech_Male> follow you online. <Silence> The best <Speech_Male> way would <Speech_Male> <Advertisement> be either my Google <Speech_Male> scholar page <Speech_Male> or <Speech_Male> my web page. <Speech_Male> That's where I <Speech_Male> put up I typically <Speech_Male> tend to put up any of <Speech_Male> my new papers, <Speech_Male> preprints directly <Speech_Male> online either on <Speech_Male> archive or SSR. And <Speech_Male> they come up <Speech_Male> on Google <Speech_Male> scholar or <Speech_Male> my web page <Speech_Male> very soon after <Speech_Male> the paper is up <Speech_Male> somewhere. So <Speech_Music_Male> that would be a great <Speech_Music_Male> spot to follow my <Speech_Music_Male> work. Yeah, we'll have <Speech_Music_Male> links to both in the show <Speech_Male> notes. Thank you. <Speech_Male> Well, thank you again for <Speech_Male> <Advertisement> taking the time to come on <Speech_Male> <Advertisement> and talk about <SpeakerChange> your work. <Speech_Music_Male> Thank you, Carl. <Speech_Music_Male> <Advertisement> Thank you. It was a pleasure <Speech_Male> being here. <Speech_Music_Male> <Advertisement> <SpeakerChange> <Speech_Male> That concludes this <Speech_Male> <Advertisement> installment of <Speech_Male> <Advertisement> data skeptic <Speech_Male> <Advertisement> ad tech. <Speech_Male> <Advertisement> Join us again next <Speech_Male> <Advertisement> week when we start <Speech_Male> <Advertisement> asking a few questions <Speech_Male> <Advertisement> about smart <Speech_Male> <Advertisement> speakers. <SpeakerChange> <Speech_Male> The smart speakers <Speech_Music_Male> are now being <Speech_Male> <Advertisement> used by millions of people <Speech_Male> and <Speech_Male> <Advertisement> we do not understand how <Speech_Male> they work, like <Speech_Music_Male> what data do they collect, <Speech_Music_Male> how that is <Speech_Music_Male> used and <Speech_Music_Male> whether that is shared with <Speech_Music_Male> anyone else.
Data Skeptic
"adword" Discussed on Data Skeptic
"Will improve on the performance of greedy. So randomization is really this tool here that I think allows you to do better. Even though you don't know the budgets. And you've touched on this a little bit, but I'm curious if we could explore the commercial applications or the settings in which this is a more appropriate way of managing things, like in AdWords, you're required to give a budget. So I guess people are used to it, and they can just do it. But there's got to be other areas that it's truly the case that a budgets unknown out priority. Do you have any examples of those? Yeah. So in the ad setting, some potential application. Again, there will be many challenges to something like this could be that if these budgets are being decided in an automated fashion by the platform, let's say, which is managing many different campaigns for an advertiser. There may be some potential for not fixing the budget for the day a priori, but adjusting it in real time. For the benefit of both the advertiser and the platform. Of course, some changing something like this which is so fundamental to search at platforms may have a lot of other challenges. So I don't know if that would necessarily in the end lead to a better bottom line. Other places where sort of not knowing these budgets a priority could be useful. For instance, if you are in a setting where there are two types of traffic, one type of traffic that is in some sense coming offline into the store, and then there's the online traffic. It turns out that this is again a case where you really want to focus on the online budget. So you want to subtract away all the offline traffic and want to focus on the remaining budget because that's better in terms of allocation. But you don't know how much the offline traffic will be what proportion of budget will it take. So if you have a budget oblivious algorithm that's doing the allocations online, it's sort of just bypasses the problem completely. Work like yours, maybe motivating a change in the way existing systems work. I know these are a system gets pretty entrenched once it's launched, but are there any theoretical motivations that would suggest we migrate to a system like this? That's a good question. So given that there are many moving parts, there are small examples that one can at least think about to say hypothetically that this could be useful. Going back, I guess, to the ad setting where a platform allowed real-time budget adjustments. It might be able to, for instance, figure out. So there's a platform that's managing two different campaigns for some organization, let's say, if it realizes that for one of these campaigns, this is an interesting day. So there are a lot of queries coming in for that kind of campaign, but not so much for the other one. It could rebalance the budget during the day and end. This could have useful lesson and could lead to a better allocation overall. Better usage of budget for the organization. But having said that, I guess, yeah, there'll be other things that one would really need to think hard about if you want more overhauling the system. What I can say is that the kind of algorithm this randomized algorithm that does do better than naive greedy strategies, even without knowing budgets, is very, very scalable. So it's not an issue of implementing it. It's easy to implement, but whether after looking at all the moving parts, it would really need to some improvement, that's something that I think, unless what really tries it out and really understands how everything else would change. Yeah, it would be unclear if it would improve the bottom line. Well, I'm curious, what's next for you? Is this part in a series of research things? Is this an offshoot or how does it fit into some of the things you're working on now? This problem and this kind of uncertainty in future. This is something that actually is inherent in many different applications. So there is this common underlying, let's say, meta problem, which is about how do you allocate resources when the demand for resources and sequential and coming in an online fashion. So that's an area that I'm very much interested in the area of online resource allocation. Yeah, I have other works in this area at the moment. Not necessarily perhaps alluded to ads. But with the same high level goal that you want to try to figure out, interesting ways of making allocations when you have uncertainty in the future and maybe other complications due to new practical applications. And you want to design an algorithm that are scalable and also do well in a theoretical case manner. Are there any other mechanisms we might add to the auction platform that could allow either the advertiser to send us more information for us to infer something, but additional techniques or rules of the system that could allow us to better optimize? So at least in the space of what we were discussing, that there's this uncertainty in future and therefore you modify the bits or shape them a little bit. There are, I believe, many different techniques, two of them are called ones threatening the other is bid scale or pacing or bid shading. Both of them have connections and in some sense, we're really inspired by some of the classical work on this bid shading idea that we talked about. Throttling, for example, is a technique that tries to for each auction. So any time a query comes, and there's the auction, will limit participation. So it will only randomly pick a subset of ads and only allow those to participate. The way this is connected to the algorithm that we talked about is that that algorithm looks at remaining budgets and if something is low on budget, it really reduces its bit. So throt is a more extreme version, which will just say that, okay, either you are in or you're out. So it's like a binary version of that idea of reducing the bids based on remaining budgets. Another idea that's also used to manage participation, what's called bit scaling or pacing and there the idea is that will reduce the bits, which is very similar to the algorithm.
Data Skeptic
"adword" Discussed on Data Skeptic
"Approach this relaxed assumption? So one could still continue to do this kind of greedy thing, which says that I'll just pick the ad that has the highest bid and show it, if it turns out that you ran out of budget at this point, you will just get this notification. Okay, you've run out of budget. And it turns out that that's not too damaging. You can still get 50% with this grid approach, which completely ignores the budgets. Other approaches that maybe use some kind of learning, maybe they could be applied in some way, but then the key question really becomes this budget that you don't know, how is it wearing? Why is it that you don't know it? Maybe there isn't a natural way to learn it anyway. The approach that I take in the paper is to say that, okay, I know nothing about these budgets because they could be part of some complicated real-time optimization problem. The could still do better than grade 8 turns out, but by randomizing our decisions. So so far, the algorithms that I briefly mentioned greedy and the other bit shading algorithms, they are deterministic, meaning that if you give some input to the algorithm, the outputs will be the same each time. Randomized algorithms, this is not the case. So I can give the same input to our Adam as algorithm. It might give me different outputs. So at different points of time, because it's using randomness in its decisions. So this randomized algorithm that can still do better than greedy even without any knowledge of budgets will still follow this high level idea of shading bits, but it will shape bits in a random way. And if you share bids in the right random, where you can do better than greedy. This seems the onset very surprising to me because if they're just random budgets, it seems like they wouldn't correlate with the true budget. So there's no information there to work with. How does the randomization how can I exploit that mathematically? The bits, the budgets are unknown and the bits are not random. Or something that the advertiser adds will come with the bits for each query. And now we still rely on the bits, but perturb them randomly. So if you think back again to this idea from a previous seminal work that you shared the bits as a function of the remaining inventory, if I have low inventory, I will try to protect that add and reduce its bit. So now that we don't know the budgets, we will do the shading randomly. So I will reduce each bit by some random amounts. But I still sort of reduce them in a multiplicative way. So let's say if I have a bit that stand and another bit that's 5, I'll shade the first bit randomly. So I'll multiply that ten by some suitable random number multiply the 5 by another independently chosen random number. But the standard phi, I still have this factor of two difference between these two shaded bits. Thanks to the initial bits being different. So it's not purely random. It's randomness in proportion to the bits, so to speak. Do we then assume that the budgets are proportional to these randomized bids or how do we get to what a budget would be? So budgets can be arbitrary. We assume in this setup, which is typically also, I believe, true in practice that the budget of an ad is much larger than any individual bit. In fact, many times any individual pick. So the total budget we still assume, even though we don't know it, is let's say hundred times just to give you a numerical example. But I don't know. It could be 200 times, it could be a thousand times. Budgets are arbitrary. They are just much larger than individual bits. And I do know the bits. I don't know the budgets. Gotcha. So then how you also don't necessarily know the real queries that are coming in, how can you evaluate a hypothetical algorithm like this one? So that's where I guess when at least we are analyzing the performance in theory, we have this very successful and popular idea which is to consider the performance in the worst case. I don't necessarily know what kinds of queries are real system might get, but I can figure out how this algorithm will do if there was an adversary out there really trying to make sure that I do terrible a terrible job at allocation. So I can look at this worst case and try to show that my algorithm will have such and such a performance, even in the worst case scenario. So that's how we sort of come up with these sort of these numbers that 50% or 63%, or in case of this randomized algorithm, something like 52.2%. Gotcha, could you give an example of that adversarial behavior if I'm an advertiser who has bad intentions? What sorts of sneaky things might I do? So it's actually less about I would say that advertiser is adversarial, but it's more about that the future of which I don't know could be adversarial from the point of view of my algorithm. So if I could go back to my to do the small example where I said that let's say we have two ads, one is a pizza store and the other is a delivery service. If you think about this scenario where if you get the first query, first query could go to either of these ads, both ads like the first query. But the second query which I don't know at the moment only goes to one of the ads. When you're trying to solve the allocation problem, you just see the first query, right? And you need to make a decision right now. And you need to pick one of the two ads to show, well, you don't know which one to pick, because you don't know the future at all. So you'll pick one of them. Let's say, and you end up picking the wrong one, because it turns out that next query is in some sense that we're serial. It's a query that will only go to the ad that you have already used the budget for. So you could only show one ad even though you got two queries in this scenario where the next query became more picky. Ah, it's like perpetual bad luck. Right. Yes. And it turns out that some of the worst instances in these kinds of problems end up being once were queries are increasingly picky over time. And you never know at a given moment, which one is really going to go off trend for future queries. So you do your best to figure out and balance all the ads. I was wondering if we could return to some of the efficiency calculations what you calculated is doing better than the greedy strategy, though admittedly not as good as the approach that uses bid shading when we knew the true budgets. Of course, we're more blinded in this situation. We couldn't possibly achieve the same results. But where do our results come from? How is it that even lacking the budget information is there some core kernel of truth of what we take advantage of to beat the greedy algorithm? So I guess the first thing is just a caveat that at the moment it's actually an open problem to figure out that if you can do as well as the deterministic bit sharing approach, even without knowing budget. So even I don't know the answer to whether you can do as well or whether you can show that you can't do as well. So coming back to your main question about how where does this however small gain on top of greedy actually come from? So I think that's really the power of randomization. So in fact, one can show that if you use deterministic algorithms and you don't know budgets, you can not beat the performance of greedy. Now the reason that's interesting is because remember that this classic approach of bit cheating is a deterministic approach, which gets you 63% and has made, I believe, significant impact on practice. When you don't know budgets, you actually can not use not just that approach, but no other deterministic approach will improve on the performance of greedy. So randomization is really this tool here that I think allows you to do better. Even though you don't know the budgets.
Data Skeptic
"adword" Discussed on Data Skeptic
"How do you approach this relaxed assumption? Thanks to our annual sponsor, waits and biases. The developer first ML platform weights and biases is an enterprise grade platform that gives you experiment tracking, dataset versioning, and model management. Their solution integrates with TensorFlow, SK learn, XGBoost, hugging face, and literally every other framework I could think of. Integrating with your model training process takes just a few lines of code. And there's even a nice live notebook on their site that will show you how to get it going. Once your machine learning code base or notebook is wired up, you've just unlocked the next level of productivity. The weights and biases platform collects aggregates and beautifully displays all of your key metrics and telemetry, so you can track your model training in real time and compare different runs. Have you ever had to screenshot a matplotlib craft that you drop in an email or slack to show your boss the results of your work? That's like serving a Gourmet dinner in a cardboard box. With weights and biases, all of your collaborators can see the results of different runs with an idealized interactive suite of tools. You no longer have to be the single source of truth for your whole organization about how the model is doing. If your honor leading a growing team of machine learning practitioners, you need to investigate if weights and biases is the right solution for you. Get started at W and B dot me slash data skeptic and mention us when you request a demo. Hi, I'm another one of those podcast ads that you've managed to tune out so effectively. It's okay. I get it. Listening to podcasts comes with some degree of ad fatigue. But today, you need to listen up because I'm here to tell you about add fatigues sinister cousin, alert fatigue. Did you know the average company receives 11,000 security alerts each day? How can a security team stay on top of the most critical threat risks when they can't find the signal on all the noise?
Data Skeptic
"adword" Discussed on Data Skeptic
"So it makes sense that we didn't do the derivation, but to say that 50% is the efficiency for greedy. It's intuitively appealing. I can trust that there's a mathematical proof there that listeners can find in the paper. And then obviously there's a ceiling here. If I had perfect knowledge of the future, I could come up with the optimal solution because I know what's going to happen in advance. I can do no better than that, but this new approach with proportional bidding to the available budget left, where does that put us between this lower bound of greedy and this theoretical upper bound of seeing the future? So the approach that I was mentioning, which really looks at the remaining budget, that's actually sort of a classical approach from 2005, let's say that basically gets you one minus one over here, which is like 63% in terms of the theoretical performance. So that's the best possible. So you can be very, very smart. You can figure out many, many complicated strategies, none will do better than this a bid shading approach in the worst case. Now the approach that I have in my paper really drives to relax one of the central assumptions in this setup and then ask the question again, can we do better than the naive greedy approach? Right, so we already have kind of a hard problem. It was enough to say, I know all the bids and I know the budgets, but I just don't know the future. How can we make this more difficult? Problems more difficult, I guess, in some sense, let's say you know even less. So what if you didn't even know the budgets? So all you knew was that, okay, you will do your allocation at some point. If you run out of budget, which you don't know, you will just get all of a sudden notification. Okay, this ad is out. It's budget has run out. But until you get to that point for any ads, you won't have any idea what the remaining budget is. So when you don't have this knowledge of budget a priority, the problem becomes interesting. I mean, I don't have an intuitive approach like reinforcement learning comes to mind, but it's not obvious what's the right methodology. How do you approach this relaxed assumption? Thanks to our annual sponsor, waits and biases. The developer first ML platform
Data Skeptic
"adword" Discussed on Data Skeptic
"So yeah, this seems pretty challenging. I don't want to be overly reductive, but I'm thinking of advertisements for umbrellas. And it rains, maybe there's more searches and more purchases and Google didn't necessarily know what was going to rain. So if all the ad budgets were spent earlier on, there would be a loss of opportunity there. Is that too simplistic of a use case? Yeah, that certainly one of the ways this thing arises. And I can give you another example. So what you pointed out, Kyle was one way in which this kind of an uncertainty is important that you want to try to spread your budget and you don't really know when it will be consumed. So to speak, or when's the best time to consume it. Another example, which captures this sort of inherent heterogeneity in queries as the following. Let's say that there are two ads. So there is a pizza store that wants to give ads on Google and then there is a delivery service that wants to give ads. Both of them have, let's say, in this stylistic example, a budget of just one ad. So there will just be $1 in this hypothetical scenario, both of them. There is a query coming in. The first query is somebody searches pizza near me. Now both the speeds are storefront and the delivery service are interested in this ad in this query. So they'll make their bit watch Google doesn't know at the moment is that the next query may be grocery deliveries. Their delivery service is interested in that query, but the Pisa storefront isn't. So if I knew at the moment that I'm going to next also get a query on grocery delivery, I could actually figure out that the right thing, maybe to do is to show for the pizza near me query, the storefront, because then I can use the budget of the delivery ad and show it to the grocery delivery query. Since I don't know that in the future, I'm going to say grocery delivery, how do I go about still making a clever decision here? So this is something that is challenging in this problem. Yeah, I think there's maybe two broad ways to look at it. One is you can try and do a lot of forecasting, guess about what queries are coming up and this sort of thing. Otherwise you could maybe just embrace some approach and say, here is my algorithm and I hope it has nice properties. Do you have any insight into the way it's actually implemented? My understanding is that both of these ideas are kind of hybridized. So the forecasting approach leads to months that at least in academia, we think of as sort of stochastic arrival models where I have some distributional knowledge about what future queries will look like. I don't know exactly what they will be, but I have a distribution. And I can use this knowledge to make better allocation decisions right now. And the other approach says that, well, what if queries are extremely hard to predict some major event happens and then that completely changes and forecasts are not valid anymore. So there's the adversarial approach that will say that I have no clue what's going to happen in the future and tries to develop algorithms that will do well even in such scenarios. And what I've been told is that these real systems at search engines try to really do both these approaches. They try to smartly figure out perhaps which is the right approach at a given time point and follow the suggestion of the respective approach. And do you have any anecdotal knowledge or even your own instincts about how good these models are? Fairly loose or can we make good predictions about this sort of stochastic system? At least at a high level, my understanding is that in ads kind of spaces, predictions can be hard. Though they do make predictions and do get a lot of useful information out of it. There are other domains which also share the same kind of similar mathematical structure like reservations on airline systems where predictions are perhaps a little bit easier than in the ad space. So both these approaches I think have seen use and the typical use cases that you try to figure out the right way to combine these approaches. And what are some of the general algorithmic approaches? Whether it's the naive first way people try, what are some of the general sorting ranking decision methods? I guess the most natural, maybe one can call it the naive way that I think you would come up with if you really sat down. I gave you the model very clearly and you were thinking about what you want to do would be what's called the greedy approach. So I remember we were thinking about just allocation and we are abstracted away all the pricing details. So when there is a query, we figure out which ads are relevant for the query. That's based on the keyword being one of the targeted keywords of the ads and so on. And then these ads are all going to make bits. So the video approach will just say pick or show the ad that has the maximum bit as long as there is still some budget of that ad remaining. That's it. So each time you get a query, look at all the bits, pick the ad that has the highest bit subject to budget being available for that ad and show that. This approach is perhaps unsurprisingly not the best approach, at least in terms of analysis, theoretical analysis, trying to understand the performance guarantee of this algorithm, one can show that this algorithm would only in some sense get you 50% in the worst case of what's you might get if you knew all the queries a priority. So there is this information gap that you have, you don't know the future and this greedy strategy gets you still 50%. One can do better than this. And that was one of the seminal results in this area. So one of the strategies that can do better than this and achieve the best possible theoretical performance does something very intuitive. So what it will say is, well, I want to be greedy, but at the same time, I want to look at how much budget is remaining in a more careful way. So if I have two potential ads, I don't just want to look at their bids for a query. I also want to look at what fraction of the budget remains for each ad and for an ad for which there is very little fraction of budget remaining, I'll share a little bit. So I'll compute these shaded bits or reduced bits for each ad and then pick the ad that has the highest shared it or reduced it and be sharing happens in a very careful way. It's a function of the remaining budget of the respective ad. And it turns out that if you shared in the right way, you can actually do substantially better than greedy, even in the worst case. This kind of an idea of sharing or considering these reduced bits has, as far as I understand, made important sort of contribution, thinking about this idea, led to some of these high level ideas like throttling and bid shading that are, I think, used in modern day search ads.
Techmeme Ride Home
"adword" Discussed on Techmeme Ride Home
"Tech memes ride home podcast recently wrote a book called how the Internet happened from netscape to the iPhone. Brian's got a history of working in the tech industry and startups. He'll talk more about that, but essentially amassed a whole lot of knowledge about the history of the Internet and wrote this book in his own words because no one else had. And we'll share a lot of interesting things that he learned along the way with us. I'll also just note that there are books for sale as usual with these things at the subsidized price for Googlers. So feel free to pick one up on your way out. And with that, Brian McCullough. Thank you, Jimmy. I'm going to stand up here so that I can quote from the book once or twice. Thank you all for coming. And thanks Google for having me come. As Jimmy said, I'm actually not a historian or a writer. This is my first book. I'm a three time web founder. Actually, funny enough, I've said this on many of my podcasts. My first company, I found it in 99. I've said many times was 100% built on AdWords. Back in the day when it was still useful and not filled with spam. But in my memory, the day that y'all opened it up for the self serve thing is the day that I got on AdWords and got I can remember paying 8 cents a click for the keyword resume and things like that. It was nutty times, but so the reason that I did this book, my second startup was 2002, my third, 2005. It just sort of bothered me that there's excellent books, of course, about the history of the arpanet and things like that. But the actual Internet going mainstream and entering normal people's lives, no one had done. And I just got tired of no one writing that book. So like any other startup I've done, I was like, well, someone's going to do that. Why not me? The dumb startup mentality. So the reason I give you that background is it's also sort of my career. It's what I lived. So I thought it would be easy to do. And so what I want to do in this talk is tell you some of the fun things that I learned in doing the research. It was 5 years of writing, essentially. And I did actual go to the library and find defunct magazines and things like real research. Also, by the way, it also because I'm a web guy and I'm used to instant gratification over the course of the writing I launched the Internet history podcast. Which there's almost 200 episodes now, which was me doing interviews for primary sources. I just had just this week Ken Norton of GV. Last month we had Matt cutts. I've had Charlie ayers on. So there's plenty of Googlers that have been on the show if you want to check that out. But the reason I give you that background is like I said, I just want to go through and sort of chronologically and tell you some of the things that even though I thought I knew all this stuff because I lived it that were sort of surprising to me. What has been surprising to me as I do media hits for this book is they almost can't help themselves. The first question, especially on radio is always, I thought Al Gore invented the Internet. So I have come up fortunately, it's in the book. I have a pat answer for that. In 1992, or actually 1991, there was a bill passed called the high performance computing act, which was a colloquially known as the gore Bill. And it funded a bunch of supercomputer facilities around the country. One of which was the NCSA, which is where mosaic, the mosaic web browser came out of, which was where Mark andreessen was going to school, which is where all the mosaic kids that would later found netscape all went to school. And so as I say in the book, essentially when they were being paid $6 an hour to code in the basement of the NCSA at the University of Illinois, they were being paid by Al Gore's gore Bill. So it's not exactly apples to apples, but in a way you could say that that played a major role in the Internet going mainstream because as I make the argument in the book, netscape was the first true dot com or web company as we would think of it today. So let's begin there, actually, because there's a million reasons why I think netscape was the Big Bang for the modern era, as I call it in the book. Not the least of which is the culture. When I started researching, the reason that I threw it up as a podcast is I worked my way through the entire mosaic engineering team that went out to California to found netscape. Sans Mark andreessen, who, even though we talk on Twitter, refuses to come on the show. But the thing that I found fascinating was this idea of startup culture. And how a lot of it was born by the media attention that netscape got when it IPO in 1995, which we'll get to in a second. To a person, all of the mosaic slash netscape engineers, almost unbidden by me, wanted to make the point that they're like, this sort of work hard, sleep under your desk, party hard, sort of startup culture that we didn't do that. The media did that to us..
WLS-AM 890
"adword" Discussed on WLS-AM 890
"Julie I didn't see all of it last night I'm just going to be straight with you I saw some clips I did my homework on it this morning I was watching the hockey game the thing bombed in the ratings the numbers just came out America's largely moved on But Tom Massey and others have asked some pretty simple questions of people like Merrick Garland very simple questions like hey were there any federal assets or federal agents involved in this whole planning or operational thing You call the insurrection on January 6th Don't you find these strange they can never really answer that question They can And it's intentional and last night instead of showing clips of interviews with FBI director Christopher wray House speaker Nancy Pelosi who was responsible for security and the capital that day Instead of seeing all the records and documentations between Pelosi's office D.C. mayor muriel Bowser's office capitol police the FBI instead of seeing legitimate records and information and interviews with the key players that day we saw nothing but theater we saw capitol police officer once again giving this overdramatic account of what happened to her that they should she's been knocked downstairs absolutely not But even her testimony officer AdWords conveniently left off the name that we still can't get an answer for Dan and that's right apps She was right there when apps was there They added it out the footage of him She did not mention his name We still can't get an answer about that but more importantly who were the FBI undercover agents and informants that day why was the capital left intentionally unsecured Where are all the FBI records Where are those records Where's the surveillance video Where's the pipe bomber What Can you I
The Dan Bongino Show
Julie Kelly: Jan. 6 Hearing Was Nothing but Political Theater
"Julie I didn't see all of it last night I'm just going to be straight with you I saw some clips I did my homework on it this morning I was watching the hockey game the thing bombed in the ratings the numbers just came out America's largely moved on But Tom Massey and others have asked some pretty simple questions of people like Merrick Garland very simple questions like hey were there any federal assets or federal agents involved in this whole planning or operational thing You call the insurrection on January 6th Don't you find these strange they can never really answer that question They can And it's intentional and last night instead of showing clips of interviews with FBI director Christopher wray House speaker Nancy Pelosi who was responsible for security and the capital that day Instead of seeing all the records and documentations between Pelosi's office D.C. mayor muriel Bowser's office capitol police the FBI instead of seeing legitimate records and information and interviews with the key players that day we saw nothing but theater we saw capitol police officer once again giving this overdramatic account of what happened to her that they should she's been knocked downstairs absolutely not But even her testimony officer AdWords conveniently left off the name that we still can't get an answer for Dan and that's right apps She was right there when apps was there They added it out the footage of him She did not mention his name We still can't get an answer about that but more importantly who were the FBI undercover agents and informants that day why was the capital left intentionally unsecured Where are all the FBI records Where are those records Where's the surveillance video Where's the pipe bomber What Can you I
WNYC 93.9 FM
"adword" Discussed on WNYC 93.9 FM
"Control freak person Part of that control is practical It makes her better at her job But she also says control is what keeps her sane Focused The only time she lets her emotions get to her is in her first minute on a new scene I have a fear before every mission I think about it and what the good happened what could happen to her to her team to the people she's treating Every time no matter how many times she does this her legs shake So many are nervous But those nerves are short lived She centers After this minute just yes Every sink could happen but I'm here and that's what I'm doing So she takes a deep breath her legs stop shaking And in she goes To do her job A list of AdWords and PR news Kyiv Ukraine.
The Charlie Kirk Show
"adword" Discussed on The Charlie Kirk Show
"In your confusion. That's pretty good. I am a pro life warrior. I'm also a known domestic terrorist in my community. I did pull my kids from school. And at home schooling them in Abby, you just took the words out of my mouth and you said that. And I've been trying to say to my Christian friends that even if you pull your kids from school and even if you put them in the private schools, I equate it to the same thing as defending all those lives if the abortion clinic, the us parents still need to go to the school board meetings and defend those children that are helpless. To stand up for them. But my question to you, Abby is, with being a pro life warrior in going to the abortion clinics in going to Planned Parenthood, we're able to stop them from getting the abortion pill there, and we're able to decide what counseling there. But I'm concerned of what you just shared today because I was unaware of that that they were able to do these appointments online and to be able to get these medications from the other countries. Is there anything that we can do or anything that you know of that I can take back to my ministry to be able to combat that in any way? How do we reach those women that are getting the medication in that form? Okay, so yeah, I mean, because this is a new battle because we're on a new front here with the medication abortion and the online accessibility of medication abortion. The only way that we can fight this is through the Internet. So we can fight this through Google AdWords. We can fight this through geofencing. So that takes a lot of money, okay? So pregnancy centers are really our best defense in this regard. So pregnancy, I don't know if you're do you work with your local pregnancy center at all? So what I would do is I would try to partner up with them. I would find out what they're doing. What are they doing with Google AdWords? Do they have an active Google AdWord campaign? How much are they funding it? Generally, to have a successful Google Ad camp and this is just FYI to have a successful Google AdWord campaign to fight medication abortion. Online medication abortion, they need to be paying at least about 30 to $40,000 a year. Okay? And that's minimum what they need to be paying to fight this. It's a lot of money. It's expensive, but lives are on the line. And this is the only way that we can fight it. You know, I mean, this is not, yes, we've got to go out to the clinics because yes, we've got to still be there for those surgical abortion patients. But over 50% are now receiving abortions online. So this is just a new battle. This is a new frontier. And we're going to have to fight it with money. And the abortion industry, they've got unlimited resources coming from Warren Buffett. And so we've got to and we have the money. We have the money. We just have to pull it. We have to utilize it. And we have to be strategic with how we use it. We only have a few more minutes, so maybe two quick questions and answers. Here, this is my first time going to a conference. You guys are all amazing. I have two burning questions, but I know we can only do one. So my first my one burning question is like, I go to a community college and this is my first year there and I've been homeschooled my whole life. And my question is, how can I reach how can you reach out to people and even teachers who are like really, very liberal and have different opinions than you do in like, how can you, how can I contradict to what they have to say to help them to come to Christ? You know, I think this goes in line with the question that I was answering earlier. I was talking about how leading by example and standing firm in your truth and sending farm in truth and in righteousness is sometimes enough. And I think one of the things that we sometimes get wrong is expecting immediate results. We want people. It's the truth. Look at it one plus one equals two. How can you not see that? And we have to realize that what we're a combating is an incredible sinister, well funded global operation propaganda machine. This is huge. You're combating a brainwash machine that has been decades in the making. It's information war. There's so much coming at them that they are being told lies over and over and over again. And sometimes telling people the truth is not enough to instantly transform them, but give it time. My transformation story obviously didn't happen overnight. I believe the lies because I was a part of the education system. It starts so early from the second that you step into the public school system. Fortunately for you, you didn't step into the public school system. So they've had 18 years of being told to lie and telling them the truth is not necessarily going to be immediate. But I think standing firm in the truth and leading by example, as I said earlier and showing people that the ways that they're invested are not actually bringing them happiness, planting seeds, sometimes is enough. And so I think I've sort of reset my expectations. And I think when I'm in every room, every contradictory room, rather, every crunch contrarian room, how many seeds can I plant today? You may be planting a lot of seeds. Don't get discouraged by the fact that you're not seeing the full plant that's supposed to sprout out. Your wallet. Ladies and gentlemen, I am sorry to say this is the part of the evening where everybody gets to hate the MC because I get to give the bad news that our time has come to an end. Can we have one more great big round of applause for all thank you guys speakers. Thanks so much for listening to everybody email us your.
WABE 90.1 FM
"adword" Discussed on WABE 90.1 FM
"Real conflict And until that happens I think that there are going to be wound that won't be healed I feel always uncomfortable about talking about the future but I'm pretty sure that eta or anything similar to it has no place in our future because 9 40 years of violence they have not achieved one political goal And I think we are all aware at least within basque society that violence was never the way Even people who supported them in the past are recognizing now that that was in their way And that it has caused too much suffering So I don't see in the future this happening again But we still have a lot of work to do I've been speaking with AdWords portela a basque writer and journalist and Tamara muru tego who was executive director of the nonprofit great mountain forest in Connecticut Thank you both so much Thank you so much Sarah It's been an honor to be here with all of you Thank you so much set up India reached out to the guardia Seville for comment on the allegation that Esteban muru teguina was subject to physical abuse while detained but we did not receive a response by airtime For 29 is our time and you're listening to all things considered on 90.1.
Voices of Search by Searchmetrics
"adword" Discussed on Voices of Search by Searchmetrics
"You can contact. Her on. twitter handle is lil- online l. i. l. l. o. n. l. i n. e. Or you could visit her company's website which is search metrics dot com a special thanks to search metrics for sponsoring this podcast. Don't forget about their new twenty twenty one offer where you can get twenty percents off of search metrics sweet and supported services until you grow your website by twenty one percent to check out the details of their twenty percents off offer go to search metrics dot co slash twenty twenty one offer and especial thanks to address for sponsoring this podcast monitoring. Your website used to require multiple expensive tools. But that's not the case anymore. Thanks to h rifts because they just launched their h. refs webmaster tools product which monitors your seo health helps you keep track of your back links and gives you the insight into what keywords performing for free so check out. Eight trips webmaster tools at eight dot com slash. Awt that's eight a. h. r. e. s. dot com slash a. w. t. Just one more lincoln. Our show notes. I'd like to tell you about if you didn't have a chance to take notes while you were listening to this podcast head over to voices of search dot com where we have summaries of all of our episodes contact information for our guests. You can also send us.
Voices of Search by Searchmetrics
"adword" Discussed on Voices of Search by Searchmetrics
"Turn your ads off and on and see what happens with the amount of traffic that you're driving and you know i think a lot of marketers are thinking about both channels at a top down view of like well. I'm gonna turn paid search off to see what the overall lift is. You should be doing that on a keyword by keyword basis and getting more granular. And you know some of this becomes a math game. Your conversion rates on your paid search ads are going to be you know one two three percent. Hopefully you're organic conversions. Depending on your placements can be significantly higher. And so there's a calculus here that goes into. What do i have to pay for the placement. What's the click through rate. What's buy organic placement to figure out which adds in which are mixed should be right and what happens. If pretend the solved and saying the result last question i have for you when it comes down to brand keywords. This every company. I've ever worked on always been a debate of should we buy our brand term somewhere that you're likely going to be the first result. Do you have an opinion on whether you should buy your brand terms through. Sem knowing that you have and has the oh placement the battle right here in offense and defense so the defense is advertising on your grand name because you're competitive bidding on it narrow show up and then the offense is going ahead and doing the same to them. And you're talking about war. So if an advertiser his side to bid on your brand name they just declared war on your company and you have a choice during gangs. You're not budging. Id recommend a defence strategy issue a cycled under attack competitors of bidding on your brand name i do recommend bidding on your brand name but this also needs to be looked at on a country-by-country basis. You might find your competitor is bidding on your one country but not another one. So don't waste your money says the real problem but yes..
Voices of Search by Searchmetrics
"adword" Discussed on Voices of Search by Searchmetrics
"To get traffic and picking up traffic probably a lot easier than vice reading competitive keywords. So i might have jumped ahead a little bit here. So we're talking about competitive skuas choking back where many different. Seo's different companies working on the website. Where you're on the content to rank for keywords that have been identified speak critical for the business and these are the same keywords that generally people managing google ads spend are also bidding on at the same time. So if you're a brand new organization new company or the likely you're not at the point where your big enough to battle for those talk placements for these really competitive keywords. But for the or broad akitas you might have more of a chance. So i think he does the answer. This question really well with your background sue. And i'm remembering some of the conversations we've had with jordan about this highly recommend people go back and listen to in your earlier conversations back in twenty nine hundred as well because actually you gave some really good advice on this topic you know. It's funny. I was just about to bring one of the conversations. I had with jordan up where his advice was. When you're starting out you should use sem to test what placements are effective understand. What keywords actually convert for you and then build your seo strategies around that where it's like. Hey i know that you know. Keywords shoes doesn't work but men's black shoes is a keyword that converts at a high rate. And we get good. Roi from it. We should build our seo strategy around men's black shoes and so often sem conserve as a testing ground to influence. What you're seo's strategy should be. I completely agree. I remember to client and he is ago. We actually built occurs as assistant that combined though such data for most eight an organic and looking for gaps during and gap analysis and then also matching not with conversions by song goes organic and pay and something that also happens a lot in this industry digital mocking general is the papal managing bad spin for the seo is vice versa..
Voices of Search by Searchmetrics
"adword" Discussed on Voices of Search by Searchmetrics
"Paint ads at the top for the organic stuff at the bottom well at the end of the day. It's real estate. We're talking about prime real estate on the incident. Is those top spots so the major such detail business so this is a huge topic of course very passionate about it considering our backgrounds. And yeah there's no answer again. I seem to say this a lot. Like i kind of shade a lot on the answers but it is a complicated topic. There are a lot of factors involved in when you consider the implications of. Do you pay for ads not do. Seo not do the same time on the same keywords or not yeah there's a lot of nuance to mastering the combination of seo versus sem right. Let's just start off by talking about some of the dynamics of sem similar to paid social. You have to pay for your media where you have to pay for your placement and your bidding for a specific keyword to get your ad shown by the difference between sem and paid social. What we talked about on monday is that in sem. Someone is in an active state of search. They're looking for a product service or education. Where they're looking for knowledge products or services. That's not the case with paid. Search right there passively browsing. They don't know what they're looking for so when someone is actively searching how do you compare and contrast the ad placement as opposed to the organic placement i think about it so paying for keywords. Painful ads choice. Certain cable. it's it's all about control so you control what your ads on you controls how much you spend per click it. Control the message. So it's there and then we look at the flip side organic. There's much less control. We do everything we possibly can to optimize for certain keywords but in the end we're really in a graze on sorry advertises who have budget and compete for clicking into the auction and can optimize keywords. They're playing a very different game and the game is about control absolutely and also of course direct response during our lie. The pressure is much higher to get a result from the spend. I think that's important thing about the your ability to control. What the messages what content. You're not at the behest of google's algorithm deciding your placement right and so that allows you to pick what your messages but also allows you to show up on pages that you wouldn't necessarily be able to rank for organically to me. That is really key into the conversation of comparing and talking about how you should use. Seo and sem is you mentioned yesterday that the keywords shoes right and how you know when somebody is looking for shoes they are not specifically in buy mode. They're doing some research who knows why someone is looking for keyword shoes..
WBZ NewsRadio 1030
"adword" Discussed on WBZ NewsRadio 1030
"Trader Joe's WalMart, Sam's Club, Costco and Publix and no longer requiring vaccinated people wear a mask inside. Of course, that doesn't matter if those stores Aaron Massa Chu You said the nurses strike It's ST Vincent Hospital in blisters now entering its 10th week, the hospital threatening to hire permanent replacements. The search for a missing Connecticut woman enters its seventh day today. 30 Year old Jessica Edwards a mom because I seen Monday morning, and South Windsor, her husband says she got dressed and left in a car with a friend. Police are considering the disappearance suspicious and haven't ruled out foul play. Her sister unique. AdWords is part of the search. Jessica The kind of person I've ever met in my whole life. She is someone that would honestly go Eddie, as far as she can to make you happy. South Windsor Police say they've been in contact with state and federal authorities and asked for help in the search. Congresswoman Liz Cheney continues to speak out after she lost her leadership position in the House this past week. Speaking on a bee sees this week, Cheney compared former President Trump's claims The 2020 election was stolen Two statements by authoritarians leaders cause that kind of questioning about our process. Frankly, it's the same kinds of things that the Chinese Communist Party says about democracy that it's a failed system in America is a failed nation. Congressional Republicans voted Friday to replace Cheney with Trump that Congresswoman Elise Stefanik as House Republican Conference chair. In California demonstrators They're calling for San Diego County's district attorney to prosecute police officers. They accused off wrongful shootings. Supporters of the racial Justice Coalition of San Diego rally outside the steps of the Hall of Justice downtown,.
Todd and Don
US Navy Official Apologizes for Calling Fired Captain ‘Stupid'
"Is it George in Huntsville Texas George is going viral he was upset about the firing of that navy captain right aboard the Theodore Roosevelt yes and now a top U. S. navy official is apologizing for calling the captain behind the coronavirus Memel naive and stupid well yeah the U. S. acting navy secretary snow apologize for calling the ousted captain of the Theodore Roosevelt too naive or too stupid to be in command amid growing calls from Congress and former officers for him to resign here's a quote let me be clear I do not think that captain Brad crozier is naive nor stupid that's what Thomas moblie road on last night I think and always believe him to be the opposite of that I apologize for any confusion this choice of words may have caused I also want to apologize directly to captain crozier his family and the entire crew of the Theodore Roosevelt to repaint my remarks may have cost more than what you say what is it why did you say those words you you said it publicly as well yeah I mean it is like a what's up man you're kidding what does that mean it's probably just a gut reaction and a bit of a pandemic I mean I don't favor just give everybody a break during this time I don't know maybe we all need to be a little bit more sensitive than everybody little foreign freedom we all in the room we all need to be patient with each other right now mostly sparked a leadership crisis at the Pentagon with the audio recording surfaced of the speech he gave to the crew of the aircraft carrier which she you know such a bad thing so that the commander had to be fired for having circulated a memo calling for more help for his crew had been stricken by the corona virus outbreak on board yeah yeah the acting navy secretary's apology came soon after Donald Trump said last night he would not intervene in the situation and hinted he might reinstate the former Roosevelt commander captain Brad crozier mmhm I I don't I don't agree with that Dodge I still think it was a horrible move it was a very very bad bad judgment there was a real security risks that we had all our judgment yes he does he does a very poor job that's worse than being naive and stupid yes it is and maybe you should apologize his career as a quote his career prior to that was very good so I'm going to get involved and see exactly what's going on there because I don't want to destroy somebody for having a bad day well now see what trump said while stressing that crozier should not have circulated that memo okay but who knows what kind of call he would have made if it was actual I mean a lot of people do look at right now as war time and that that we are fighting a battle but who knows what kind of decision this whatever this guy would've made in the heat of actual combat as we know it I mean if he does this what else is he capable of doing you know what I mean it's respect cannot be taken well you're you're describing him worse than what the acting navy adwords I think it I think it's just unspeakable what he did I I really do I I I think that for so many reasons not only a safety issue but also just just common sense it tells me the guy just didn't have much common sense to know he was desperate but still it's kind of a stretch I mean this guy served this country for decades or are you are you are you starting to ease up on a little bit he's up on it now I don't know the whole story none of us do he did he did break the rules of going around the chain of command which you should have never done so you should have carried that further up the chain of command to get what he needed absolutely it revealed a few secrets that you shouldn't reveal when you're leading that kind of ship yes but he has served this country with the highest regard and I'm not gonna you know you have one bad day not referring to moblie enclosure the president added to good people they were arguing and law and I'm good I believe it or not and shuttling argument so I may have to look into this great detail and I'll try to figure out if he's able to do anything about it now in the recording of moblie speech on the board of the Roosevelt published by task and purpose military news website heckling can be heard from the crew during the speech what shows otherwise greeted with silence talking about moblie speech an apology to the crew of the ship mostly justified his decision to relieve the captain of command just last week on the announcer this four page memo requesting more helpful is cruel that it asking to be moved on to the short qualm and if it's circulating more than forty people I've seen that and it was seen as kind of a you know sidestepping the chain of command what you don't do when your you know driving and the captain of a nuclear ship like that what trump is thinking about getting involved in this and he says I don't want to see anybody you lot and then buys live record because of one band that that one bad day I don't know maybe it's realistic again unrealistic I've never been in the military never been close to it however I just kind of my opinion my view that you can't have a bad day in the
AP 24 Hour News
Governors are frustrated over medical supply shortage
"Nation's governors are expressing frustration with the scarcity of medical supplies it's very respond to the corona virus pandemic I can't tell you that over the last seventy two hours the flow of personal protective equipment PPE has improved but Louisiana governor John bel Edwards says he's told the White House it needs to be more really doing just in time deliveries right now were emptying out our warehouse every single night adwords says Louisiana's biggest need is ventilators he's ordered fourteen thousand the number however that we've received continues to be a hundred and ninety one president trump says more than nine thousand ventilators have been shipped to states from the federal stockpile the ten thousand are being held back we didn't want to give them because we don't know where the emergency this hits it hits like so fast it comes so quickly the president promises a lot more are in production they're talking about hundreds of thousands being made in a very short period of
Bloomberg Daybreak: Europe
What happens next with Brexit could make or break the UK
"I want to get to a very London topic and that is brexit I'm sure no one's tired of hearing about the ups and downs of the UK's attempts to leave the European Union the blame game has escalated now is talks between Britain and Europe seem to have reached an impasse yet again Boris Johnson is due to meet with his Irish counterpart by the end of the week but Leo Varadkar says that hashing out of brexit deal would be very difficult right now for more we go live to Dublin where Bloomberg anchor an adwords is standing by Anna what do you hearing from Ireland good morning teen nothing Caroline yeah well Lee of records you say as stressing that he sees it as hard to get a deal done because of the the the the the state of the impasse and it's hard to break that impasse he said my next week and next week masses because that is a easy leaders summit taking place in Brussels the border border issues along the Irish border that remains one of the big sticking points on on and it's a sticking point on a number of levels first lead the extent to which you'd have to have customs checks how they would work where they would be the Irish is saying they want to see any kind of costumes checking on the island of Ireland that he's a big at pots of the sticking point at the end the sticking points here the other one really to do with any vetoes at various political parties in Northern Ireland will be given on the extent of the closeness in the E. U. regulates read regime and how much that would be adopted by Northern Ireland Sir the various parts of the northern islands of political spectrum see the veto was working in the other parties favor as a for that reason it remains contentious but as you say we are expecting to see a meeting between Iraq on Boris Johnson tomorrow or Friday to trying to come up with some way of breaking this impossible to possible yeah because of cool so the problem and the fallout is being spelled out by this morning in the latest details on the bank of England's financial policy committee warning that the U. K. may face economic turmoil in and no deal breaks it all the lights Reykjavik terminal for the bank being in the same with you K. risks big economic help people if there is an ideal breaks it bots that U. K. banks as prepared as they can be for breaks it say those sort of latest news on it's coming out of the bank of England's of the U. K. perspective alone all of the S. M. but obviously you over in on it because of the budget because that that happened this week and they're all deep concerns in all at about the the full out so for the economy that would markets pricing in what's the perspective on the impact of a potentially a delay or no duplexes yeah well I just spoke to John Cronin who's that from the Irish or British Irish chambers of commerce and that he was saying you know make no mistake it is not possible for businesses to prepare entirely for no deal brexit OSes back to you the I. E. C. which is a representative of the business and employees association here and they were saying that the the the pain of a no deal brexit would fall disproportionately on certain parts of the Irish economy said the agricultural base communities the rural communities that is where the pain would face he said well then in the cities which are more reliant on the services sector which could be a little bit more resilient but there was clearly a lots of concern here and as you say we have the budget yesterday and the Irish government making that budget at entirely based around and no deal brexit will be speaking to the finance minister little bit likes one ask you more about that in terms of what the market is pricing in the Carolinas interesting that we talking about this deadline in just three weeks in a day that's how long it is until the end of October of course and actually the market is not really focused on that day market at a pound invested in particular focusing more on November and December volatility as if they're trying to look through the thirty first of October deadline perhaps they don't think that it will actually happen on that day that day that market participants are expecting
Techmeme Ride Home
Google is retiring the AdWords & DoubleClick brands in a major rebranding aimed at simplification
"So this is just a story of corporate rebranding but if you're in a certain segment of the web game this is really big news all of google's suite of advertising products will now go under one of three new ombrella 's atwood's will now be known as google ads the doubleclick for advertisers product and google analytics will now be known as the google marketing platform and doubleclick for publishers as well as the doubleclick attic change will now go under the moniker google ad manager so at words of course are those text ads that you see when you do a search on google they used to run along the side but they've been integrated in the main search results for a while now adverts was google's first major product nearly eighteen years ago the key innovation that turned on google's money geyser for the first time and made google into a juggernaut doubleclick is actually twenty two years old it was the first banner ad juggernaut back in the dot com.